Dans le cadre de la transition énergique, le déploiement de sources d’énergies ne produisant pas de gaz à effet de serre devient primordial. Bénéficiant de la surabondante énergie fournie par le Soleil, le photovoltaïque est un des éléments-clés du bouquet énergétique du futur. Le marché du photovoltaïque est actuellement dominé par les technologies à base de silicium et les meilleurs rendements de conversion dépassent les 26% avec la technologie de cellules à hétérojonction de silicium amorphe hydrogéné (a-Si:H) sur silicium monocristallin (c-Si).Le silicium amorphe hydrogéné, déposé par PECVD, permet d’obtenir une excellente passivation de la surface du substrat de silicium cristallin, et ainsi d’obtenir des tensions de circuit ouvert au-delà de 730 mV. Cependant l’a-Si:H montre une absorption parasite des photons ultraviolets, et sa faible conductivité limite la longueur de diffusion des porteurs de charge générés en son sein, limitant la performance électrique et aussi leur contribution au courant de la cellule.Pour augmenter le rendement de cette technologie, nous proposons de fabriquer et de caractériser une nouvelle structure de cellules photovoltaïques à base d'hétérojonction de phosphure de gallium (GaP) sur c-Si, déposé par dépôt en phase vapeur aux organométalliques (MOCVD). Matériau III-V, cristallin, et à énergie de bande interdite élevée (2.26 eV contre 1.6-1.9 eV pour l’a-Si:H et 1.12 eV pour le c-Si), le GaP permettrait une croissance par épitaxie sur le c-Si, une meilleure transparence face à l’a-Si:H, ainsi qu’une passivation par effet de champ repoussant les trous, porteurs de charge positive, loin de l’interface GaP/Si. Les améliorations des caractéristiques courant-tension de telles cellules avec seulement 10 nm de GaP ont précédemment montré, par simulation, une amélioration des rendements de 2% en absolu.Dans le cadre de cette thèse, nous avons étudié expérimentalement l’effet du dépôt de GaP sur le c-Si. Nous avons mis en évidence une dégradation de la durée de vie des porteurs dans le c-Si lors d’une étape de préparation de surface pour améliorer l’épitaxie du GaP, qui favoriserait la diffusion de contaminants issus de la chambre de dépôts III-V dans le substrat. Cette étape pourrait être retirée, mais elle est nécessaire pour limiter l’émergence de domaines d’antiphase, défauts cristallins liés à la nature polaire des liaisons Ga-P qui limitent aussi la durée de vie des porteurs. De plus, la durée de vie à l’interface GaP/Si est demeure inférieure à 150 µs, malgré l’hypothétique passivation par effet de champ et sans défauts cristallins.Se basant sur ces découvertes, nous avons cherché à comprendre et améliorer la passivation de l’interface GaP/Si. Des techniques d’analyses avancées ont montré la présence de traces de carbone et d’arsenic dans le GaP, accompagné de fluor à l’interface, ainsi qu’une oxydation du GaP post-épitaxie. Différentes couches de mouillage ont été testées, permettant de corréler la rugosité, la défectuosité du GaP à la durée de vie des porteurs.D’autre part, l’intégration d’étapes de décontamination du substrat (gettering) a permis avec succès de restaurer la durée de vie volumique des charges tout en maintenant le recuit de reconstruction de surface dans le procédé de fabrication. Ces étapes ont été optimisées pour minimiser leur impact sur la couche de GaP. Un cellule avec GaP déposé sans pré-recuit atteint 11.2% tandis qu’en reléguant le GaP à une couche fenêtre, une cellule GaP/(n+)c-Si/(p)c-Si a montré un rendement amélioré à 13.8% avec le recuit et les étapes de gettering.Ce travail s'appuie sur l'expertise du CEA-INES en cellules solaires à hétérojonctions et du CNRS-LTM en épitaxie et caractérisation des matériaux III/V. / In the frame of energy transition, the development of energy sources that do not generate greenhouse gases is paramount. Benefiting from the overabundant energy provided by the Sun, photovoltaics is a key element of the future energy mix. Photovoltaics market is currently led by the silicon-based technologies, and best conversion efficiencies exceed 26% with the heterojunction solar cells technology with hydrogenated amorphous silicon (a-Si:H) on monocrystalline silicon (c-Si).Hydrogenated amorphous silicon, deposited by PECVD, enables high surface passivation of crystalline silicon, and to reach over 730 mV of open-circuit voltage. However, the parasitic absorption in the Ultra Violet region limits photon collection, and its low conductivity limits the diffusion length of charge carriers it generates, limiting the electrical performance and their contributions to the cell current.To enhance the efficiency of this technology, we propose to fabricate and characterise a new structure of photovoltaic solar cells based on heterojunction of gallium phosphide on crystalline silicon, made by metalorganic chemical vapour deposition (MOCVD). This crystalline III-V material, with high bandgap energy (2.26 eV vs 1.6-1.9 for a-Si:H and 1.12 eV for c-Si), allows its pseudomorphic epitaxy on silicon, with higher transparency vs a-Si:H along with field effect passivation that repels the holes, positive charge carriers, away from the GaP/Si interface. The improvement of current-voltage characteristics, with only 10-nm-thick GaP, have previously shown by simulation an absolute improvement of the efficiency by 2%.In the frame of this thesis, we have experimentally studied the effect of GaP deposition on c-Si. We have outlined a carrier lifetime degradation in c-Si during a surface preparation annealing that favours the diffusion of contaminants from the III-V MOCVD chamber into the substrate. This step could be removed, but it is required to limit the formation of antiphase domains, which are crystalline defects linked to the polarity of Ga-P bonds that also limit the carrier lifetime. Moreover, GaP/Si interface lifetime remains below 150 µs, despite the hypothetic field effect passivation and without crystalline defects.From these conclusions, we sought to understand and improve the GaP/Si interface passivation. Advanced analysis techniques have shown carbon and arsenic traces in the GaP, with fluorine at the interface, as well as post-epitaxy GaP oxidation. Different wetting layers were tested, correlating the roughness and defectivity of Gap to the carrier lifetime.Furthermore, integration of substrate decontamination steps (gettering) enables successful bulk carrier lifetime recovery while maintaining the surface reconstruction annealing in the process flow. These steps were optimised to minimise their impact the GaP layer. A solar cell with GaP deposited on unannealed silicon reached 11.2% while, making GaP a window layer in a GaP/(n+)c-Si/(p)c-Si stack produced a solar cell with 13.8% with annealing and gettering steps.This work relies on the expertise of CEA-INES on heterojunction solar cells and CNRS-LTM on the epitaxy of III-V materials and their characterisation.

In this work novel solid-state neutron detectors based on Gallium Nitride (GaN) have been produced and characterized. GaN is a radiation hard semiconductor which is commonly used in commercial optoelectronic devices. The important design consideration for producing GaN-based neutron detectors have been examined, and device simulations performed. Scintillators and p-i-n diode-type neutron detectors have been grown by metalorganic chemical vapor deposition (MOCVD) and characterized. GaN was found to be intrinsically neutron sensitive through the Nitrogen-14 (n, p) reaction. Neutron conversion layers which produce secondary ionizing radiation were also produced and evaluated. GaN scintillator response was found to scale highly linearly with nuclear reactor power, indicating that GaN-based detectors are suitable for use in the nuclear power industry. This work is the first demonstration of using GaN for neutron detection. This is a novel application for a mature semiconductor material. The results presented here provide a proof-of-concept for solid-state GaN-based neutron detectors which offer many potential advantages over the current state-of-the-art, including lower cost, lower power operation, and mechanical robustness. At present Helium-3 proportional counters are the preferred technology for neutron detection, however this isotope is extremely rare, and there is a global shortage. Meanwhile demand for neutron detectors from the nuclear power, particle physics, and homeland security sectors requires development of novel neutron detectors which are which are functional, cost-effective, and deployable.

A wide variety of group III-Nitride-based photonic and electronic devices have opened a new era in the field of semiconductor research in the past ten years. The direct and large bandgap nature, intrinsic high carrier mobility, and the capability of forming heterostructures allow them to dominate photonic and electronic device market such as light emitters, photodiodes, or high-speed/high-power electronic devices. Avalanche photodiodes (APDs) based on group III-Nitrides materials are of interest due to potential capabilities for low dark current densities, high sensitivities and high optical gains in the ultraviolet (UV) spectral region. Wide-bandgap GaN-based APDs are excellent candidates for short-wavelength photodetectors because they have the capability for cut-off wavelengths in the UV spectral region (λ < 290 nm). These intrinsically solar-blind UV APDs will not require filters to operate in the solar-blind spectral regime of λ < 290 nm. For the growth of GaN-based heteroepitaxial layers on lattice-mismatched substrates, a high density of defects is usually introduced during the growth; thereby, causing a device failure by premature microplasma, which has been a major issue for GaN-based APDs. The extensive research on epitaxial growth and optimization of Al_x Ga _1-x N (0 ≤ x ≤ 1) grown on low dislocation density native bulk III-N substrates have brought UV APDs into realization. GaN and AlGaN UV <i> p-i-n </i> APDs demonstrated first and record-high true avalanche gain of > 10,000 and 50, respectively. The large stable optical gains are attributed to the improved crystalline quality of epitaxial layers grown on low dislocation density bulk substrates. GaN <i>p-i-n </i> rectifiers have brought much research interest due to its superior physical properties. The AIN-free full-vertical GaN<i> p-i-n </i> rectifiers on<i> n </i>- type 6H-SiC substrates by employing a conducting AIGaN:Si buffer layer provides the advantages of the reduction of sidewall damage from plasma etching and lower forward resistance due to the reduction of current crowding at the bottom<i> n </i> -type layer. The AlGaN:Si nucleation layer was proven to provide excellent electrical properties while also acting as a good buffer role for subsequent GaN growth. The reverse breakdown voltage for a relatively thin 2.5 μm-thick<i> i </i>-region was found to be over -400V.

This thesis describes the development of III-Nitride materials for light emitting applications. The goals of this research were to create and optimize a green light emitting diode (LED) and laser diode (LD). Metalorganic chemical vapor deposition (MOCVD) was the technique used to grow the epitaxial structures for these devices. The active regions of III-Nitride based LEDs are composed of InₓGa₁₋ₓN, the bandgap of which can be tuned to attain the desired wavelength depending on the percent composition of Indium. An issue with this design is that the optimal growth temperature of InGaN is lower than that of GaN, making the growth temperature of the top p-layers critical to the device performance. Thus, an InGaN:Mg layer was used as the hole injection and p-contact layers for a green led, which can be grown at a lower temperature than GaN:Mg in order to maintain the integrity of the active region. However, the use of InGaN comes with its own set of drawbacks, specifically the formation of V-defects. Several methods were investigated to suppress these defects such as graded p-layers, short period supper lattices, and native GaN substrates. As a result, LEDs emitting at ~532 nm were realized. The epitaxial structure for a III-Nitride LD is more complicated than that of an LED, and so it faces many of the same technical challenges and then some. Strain engineering and defect reduction were the primary focuses of optimization in this study. Superlattice based cladding layers, native GaN substrates, InGaN waveguides, and doping optimization were all utilized to lower the probability of defect formation. This thesis reports on the realization of a 454 nm LD, with higher wavelength devices to follow the same developmental path.

The dielectric tensor of Ga2O3 in the monoclinic (β) phasewas determined by generalized spectroscopic ellipsometry in a wide spectral range from 0.5 eV to 8.5 eV as well as by density functional theory calculations combined with many-body perturbation theory including quasiparticle and excitonic effects. The dielectric tensors obtained by both methods are in excellent agreement with each other and the observed transitions in the dielectric function are assigned to the corresponding valence bands. It is shown that the off-diagonal element of the dielectric tensor reaches values up to |εxz| ≈ 0.30 and cannot be neglected. Even in the transparent spectral range where it is quite small (|εxz| < 0.02) it causes a rotation of the dielectric axes around the symmetry axis of up to 20◦.

We present experimental studies based on all three methods by which the generation of squeezed light in semiconductors has thus far been demonstrated experimentally: Fourwave mixing, multi-photon absorption and direct generation at the source. Four-wave mixing was used to generate femtosecond-pulsed quadrature squeezed light by cross-phase modulation in single-crystal hexagonal CdSe at wavelengths between 1.42 and 1.55 μm. We measured 0.4 dB squeezing (1.1 dB is inferred at the crystal) using 100 fs pulses. The wavelength and the intensity dependence, as well as variations in the local oscillator configuration were investigated. At higher intensities squeezing was shown to deteriorate owing to competing nonlinear processes. We also characterised the nonlinear optical properties of CdSe in this wavelengths range using an interferometric autocorrelator. In addition, we studied the feasibility of extending this technique to AlGaAs waveguides. The key problems are addressed and solutions are proposed. In a different experiment we used an AlGaAs waveguide to demonstrate for the first time photon-number squeezing by multi-photon absorption. By tuning the pump energy through the half bandgap energy we could effectively select two- or three-photon absorption as the dominant mechanism. Squeezing by these two mechanisms could be clearly distinguished and was found to be in good agreement with longstanding theoretical predictions. We also established the generality of the effect, by demonstrating the same mechanism in organic semiconductors, where it led to the first ever observation of squeezed light in an organic material. Finally, we present our measurements of photon-number squeezing in high-efficiency double heterojunction AlGaAs light-emitting diodes. We measured squeezing of up to 2.0 dB. In addition, we observed quantum noise correlations when several of these devices were connected in series.

There are significant differences between experiment and theoretical calculations of the electronic structure of GaSb/GaAs self-assembled quantum dots. Using a multi-band effective mass approximation it is shown that the influence of size and geometry of quantum dots has little or no effect in determining the hydrostatic strain. Furthermore, the valenceband ground state energies of the quantum dots studied are surprisingly consistent. This apparent paradox attributed to the influence of biaxial strain in shaping the heavy-hole and light-hole potentials. Consequently, it is shown that a simple, hydrostatically derived potential is insufficient to accurately describe the electronic structure of such quantum dots. In addition, using the latest experimental results measuring the conductionband offset, it has been shown that much better experimental contact may be achieved for the magnitude of the transition energies derived compared to theoretically derived transition energies. The transition energies of Si/Ge self-assembled quantum dots has also been calculated. In particular, a range of quantum dot structures have been proposed that are predicted to have an optical response in the 3-5 micron range.

Thesis (Ph.D.)--University of Delaware, 2008. / Principal faculty advisors: Valeria Gabriela Stoleru, Dept. of Materials Science & Engineering; and S. Ismat Shah, Dept. of Materials Science. Includes bibliographical references.

Thesis (PhD)--Macquarie University, School of Mathematics, Physics, Computing and Electronics, 1999. / "1998"--T.p. Includes bibliographical references.

Th.--Sci.--Toulouse 3, 1981. N°: 1001.