Der Quantenkaskadenlaser (QCL) ist ein unipolares Intersubbandbauelement dessen Funktionsweise auf Übergängen zwischen dem ersten angeregten Zustand und dem Grundzustand in einem Quantentopf (quantum well, QW) beruht. Er wurde im Jahre 1974 von Kazarinov und Suris theoretisch vorhergesagt und erstmals 1994 von Faist et al. experimentell realisiert. Das Elektron verlässt nach dem Laserübergang nicht das Leitungsband und kann somit durch ein angelegtes elektrisches Feld in die nächste aktive Zone transferiert werden, wo es wiederum einem Laserübergang untergehen kann. Schliesslich, nach einer Reihe solcher Kaskadenprozesse, emittiert ein einzelnes Elektron viele Photonen; dies definiert die hohe Quanteneffizienz der QCLs. Das Hauptproblem bei der kaskadierten Benutzung von aktiven Regionen ist ein schneller Elektronentransport zwischen den emittierenden QWs mithilfe des sogenannten Injektors. Ein schneller Transport der Ladungsträger ist notwendig um das obere Laserniveau zu populieren und das untere zu depopulieren, womit die für die stimulierte Emission notwendige Besetzungsinversion erreicht werden kann. Zur Gewährleistung des schnellen Transports im Injektor ist die Verwendung von Materialien mit einer geringen effektiven Masse naheliegend. Unter den technologisch wichtigen III-V Verbindungen besitzt InAs die geringste elektronische effektive Masse von 0.023m0 (wobei m0 die Masse des freien Elektrons ist). Die binäre Verbindung mit der nächst grösseren effektiven Masse ist GaAs mit m*=0.067m0. Bisher wurden QCLs in beiden, InAs und GaAs und weiterhin im ternären InGaAs basierten QW Materialsystem realisiert. Gegenwärtig zeigen QCLs einen hohen Grad der Reife; hohe Lichtleistung, Dauerstrichbetrieb und Betrieb bei Raumtemperatur sowie Oberflächenemission wurden erzielt. Der von den QCLs abgedeckte spektrale Bereich erstreckt sich von 3.5 Mikrometer bis zu 87 Mikrometer. Trotz des hohen Reifegrades ist der Quantenkaskadenlaser immernoch in der Entwicklung. Speziell die Erweiterung des spektralen Bereichs ist für viele Anwendungen essentiell. Enorme Fortschritte bei der Erweiterung hin zu grösseren Wellenlängen wurden in den letzten Jahren erzielt, dennoch ist der kurzwellige Rekord von 3.5 Mikrometer aus dem Jahre 1998 bisher ungebrochen. Nichtsdestotrotz besitzt der QCL auch im nahen Infrarot das Potential den konventionellen Interbandlaser zu übertreffen. Neben dem Wettstreit um Schwellströme und Ausgangsleistungen, ist aufgrund der andersartigen Physik des Laserüberganges eine verbesserte Anwendungsmöglichkeit im Bereich des schnellen optischen Schaltens zu erwarten. Die Herausforderung im Bereich der kurzwelligen QCLs liegt in der beschränkten Leitungsbanddiskontinuität (CBO) zwischen Quantentopf- und Quantenbarrierenmaterial. Um zwei gebundene elektronische Eigenzustände innerhalb der Quantentöpfe der aktiven Zone zu gewährleisten, wird eine grosse Leitungsbanddiskontinuität benötigt. Weiterhin kann nur so eine ausreichend hohe Barriere zwischen den angeregten Zuständen und dem klassischen Zustandskontinuum bei angelegtem elektrischen Feld erreicht werden. Neben der Notwendigkeit des grossen CBO sollte das Barrierenmaterial eine direkte Bandlücke aufweisen oder zumindest der angeregte Zustand in der aktiven Zone unterhalb des niedrigsten Leitungsbandes des Barrierenmaterials liegen. Mit der Einschränkung bezüglich der Gitterkonstanten von Quantentopf und -barrierenmaterial für ein koh ärentes Wachstum auf einem bestimmten Substrat, endet man bei nur einer Hand voll vielversprechender Materialkombinationen für die Anwendung in QCLs. Das grösste CBO für Materialien mit direkter Bandlücke findet man bei InGaAs/InAlAs. Wir erzielen 520 meV für die ternäre an InP gitterangepasste und 740 meV für die spannungskompensierte In(0.70)Ga(0.30)As/In(0.40)Al(0.60)As Kombination. Unter den Barrierenmaterialien mit indirekter Bandlücke ist die Kombination InAs/AlSb auf GaSb oder InAs mit 2.1 eV CBO im Gamma-valley sehr vielversprechend. Quantenkaskadenlaser basierend auf diesem Materialsystem mit Emission bei 10 Mikrometer wurden kürzlich von Ohtani and Ohno realisiert. Jedoch wurde im kurzwelligen Bereich um 4 und 3 Mikrometer in diesem System bisher nur spontane Emission beobachtet. Damit ist es bis heute ein offene Frage, welches Materialsystem tatsächlich das geeignetste für die Anwendung in kurzwelligen QCLs sein wird und ob es überhaupt möglich sein wird, ihren Wellenlängenbereich auf die Telekommunikationswellenlänge von 1.55 Mikrometer auszuweiten, was zweifellos die grösste Herausforderung darstellt. Oberflächenemission von QCLs ist bisher mittels der Aufbringung einer Rippenstruktur mit kurzer Periode auf der Oberfläche der Laserstreifen erreicht worden. Die Möglichkeit einer Polarisation in der Fläche mithilfe selbstorganisierter Quantenpunktstrukturen innerhalb der aktiven Zone ist ein aktuelles Thema innerhalb der QCL-Gemeinschaft, aber bisher noch unerreicht. Die Kombination aus feldinduzierten Minibändern aus elektronischen Zuständen in konventionellen QCLs und diskreten atomartigen Zuständen in Quantenpunkten ist eine kreative und gleichzeitig widersprüchliche Idee. Dennoch vereint dieses Thema ein gewaltiges Interesse sowohl von theoretischer als auch experimenteller Seite innerhalb der QCL-Gemeinschaft. Diese Arbeit ist der Erweiterung der Materialvielfalt für die Herstellung von Quantenkaskadenlasern gewidmet. Die Mission dieser Forschungsarbeit ist - die Grenzen im Gebrauch des spannungskompensierten Designs des klassischen InGaAs/InAlAs Materialsystems auf InP für kurzwellige Emission auszuloten; - die Möglichkeiten kurzwelliger Intersubbandemission in einer der extraordinären Materialkombinationen für die QCL-Anwendung zu erforschen: spannungskompensiertes InGaAs/InGaP auf GaAs; Die Quintessenz der gesamten Forschungsarbeit besteht in der spannungskompensierten Herangehensweise und den InGaAs enthaltenden Materialsystemen für die Anwendung in Quantenkaskadenlasern. Die Arbeit ist wie folgt strukturiert: Kapitel 1: Die vorliegende Einführung. Kapitel 2: Kurzer überblick der Eigenschaften von Intersubbandübergängen und der Grundlagen der QCL-Funktionsweise. In diesem Kapitel wird eine Einführung in die Eigenschaften von Intersubbandübergängen und den Minibandtransport gegeben. Dieses Kapitel unterstreicht den physikalischen Unterschied von Intersubbandübergängen und Transport zum Fall der Interbandübergnge und gibt eine Einführung in die vorteilhaften Eigenschaften der Intersubbandbauelemente. Weiterhin wird eine Einführung in die Physik des Quantenkaskadenlasers und eine übersicht der Designvielfalt der aktiven Zone gegeben. Im Speziellen wird auf die unterschiedlichen Strategien bei der Erzielung der Besetzungsinversion eingegangen. Kapitel 3: Experimentelles Kapitel. Das 3. Kapitel fasst die erzielten eigenen Ergebnisse innerhalb des InGaAs/InAlAs Materialsystems auf InP zusammen. Dabei konzentriert es sich auf extreme Fälle des spannungskompensierten Designs welche die Realisierung kurzwelliger übergänge zum Ziel haben. Kapitel 4: Experimentelles Kapitel. Im 4. Kapitel werden die erzielten eigenen Ergebnisse innerhalb des InGaAs/InGaP Materialsystems dargestellt. Das InGaAs/InGaP Materialsystem auf GaAs wurde unseres Wissens zuvor füür Intersubbandbauelemente weder benutzt noch vorgeschlagen. Das Kapitel beschreibt den gesamten Verlauf, beginnend mit dem Probenwachstum über grundlegende Materialstudien, bis hin zum Design der QC-Teststruktur und deren Fabrikation. Kapitel 5: Hierin wird die Zusammenfassung der erzielten eigenen Ergebnisse und daraus resultierenden Schlussfolgerungen gegeben. / Quantum cascade lasers, QCL, are unipolar intersubband devices, which work on transitions between the first excited and the ground state in quantum wells, QW. They where predicted theoretically by Kazarinov and Suris 1974, and realized experimentally for the first time by Faist et al. 1994. Electron does not leave the conduction band after the lasing transition in QCL. And therefore it can be used again in the next active region, where it can be transferred due to applied electric field. Finally, after a number of such cascade processes, single electron emits many photons, which defines a high quantum efficiency of QCLs. The key issue in use of cascaded active regions is a fast electron transport in between the emitting QWs (so called, injector region). Fast carrier transfer is needed on the one hand to effectively populate the upper lasing state in active region QW and on the other hand to quickly depopulate the lower lasing state. So that population inversion, necessary for stimulated emission, is achieved. To provide the fast transport in injector region it is likely to deal with materials with a low effective mass. Among the variety of technologically important III-V compounds InAs has the lowest electron effective mass of 0.023m0 (where m0 is the free electron mass). Next low effective mass binary material after InAs is GaAs with m*=0.067m0. Up to now QCLs are realized on both, InAs- and GaAs- as well as ternary InGaAs-based-QW material systems. Currently QCLs show a high level of maturity. High power, cw-operation and room temperature operation as well as surface emission are achieved. Spectral range, covered by QCLs, extends from 3.5 micrometer up to 87 micrometer. Despite of the high level of maturity, QCLs are still under development. In particular, extension of the spectral range of operation is likely for many applications. Tremendous progress was achieved last years in long wavelength range extension of QCLs. However, the short wavelength record of 3.5 micrometer has not been beaten since 1998. Nevertheless, QCLs has a potential to outperform conventional interband lasers also in near infrared spectral range. Apart from competition in threshold current densities and output power, QCLs are expected to be better in fast optical switching operation due to different physics of lasing transitions. The challenge of short wavelength QCLs is a limited conduction band edge offset, CBO, between the quantum well and barrier material. High CBO is needed to confine two quantized electron states in active region QW and to provide sufficient barrier between the excited state and classical continuum of states above the barrier material conduction band edge under applied electric field. More over, despite of high CBO demand, barrier should be the direct band gap material, or at least, the upper lasing state in active region should lay below the lowest conduction band valley in the barrier material. Together with restriction on the lattice constant of both, well and barrier materials, for coherent growth on a certain substrate, we end up with very few promising material combinations for QCL application. The highest CBO for direct band gap materials combination we find in InGaAs/InAlAs. We obtain 520 meV for lattice matched to InP ternaries and about 740 meV for strain-compensated In(0.70)Ga(0.30)As/In(0.40)Al(0.60)As combination. Among the indirect barrier material combinations, very promising is InAs/AlSb on GaSb or InAs with 2.1 eV CBO in gamma-valley. QCL emitting at 10 micrometer has been recently realized on this material system by Ohtani and Ohno. However, at short wavelength, 4 and 3 micrometer, only spontaneous emission is obtained in this material system up to now experimentally. So up to now, it is still an open question, which material system is going to be most suitable for short wavelength QCL application. And it is still an open question, if it is possible at all to extend the operation wavelength of QCLs to the most challenging 1.55 micrometer telecommunication wavelength. Surface emission is achieved in QCLs up to now by manufacturing of the short period grating on the top of the planar laser stripe. The possibility of in-plane polarized emission involving self organized quantum dot structures into the QCL active region is a hot topic in QCL community, but it is not achieved experimentally up to now. Combining the field induced minibands of electron states in conventional QCLs together with discrete atom-like states in QDs is a creative and at the same time contradictive idea. Nevertheless, this topic attracts a huge interest from both, theoretical and experimental, side of QCL community. This work is dedicated to make a step forward in extension of material variety used for QCL fabrication. The mission of this research is - to find out the limits of use of strain-compensated designs on classical InGaAs/InAlAs material system on InP to achieve the short wavelength generation; - to discover the possibilities of short wavelength intersubband generation in one of extraordinary material combinations for QCL application: strain-compensated InGaAs/InGaP on GaAs; The bottom line of the whole research is strain compensation approach and InGaAs containing material systems for QCL application. Present work consist of: Chapter 1: The current introduction. Chapter 2: Brief overview of intersubband transitions properties and the basics of QCL action. In the overview-chapter an introduction into the properties of intersubband transitions and miniband transport is given. This chapter underlines the difference in physics of intersubband transitions and transport comparing to the case of interband transitions; and gives an introduction into the advantageous properties of intersubband devices. This chapter gives an introduction into the quantum cascade laser physics and overview on variety of active region designs. This chapter is, specially, dedicated to point out different ways of achieving the population inversion in each QCL active region approach. Chapter 3: Experimental chapter. Third chapter describes obtained original results on InGaAs/InAlAs material system on InP during the present work. It concentrates on extreme cases of strain-compensated designs for achieving the short wavelength transitions. Chapter 4: Experimental chapter. Forth chapter describes obtained original results on InGaAs/InGaP material system. InGaAs/InGaP material system on GaAs was never before, up to our knowledge, proposed or used for intersubband devices. So, the chapter describes all the way from the sample growth issues and basic study of this material up to QC test-structure design and fabrication. Chapter 5: Here, the summary of obtained original results and conclusions are given.

Ce travail de thèse a pour but de contribuer au développement d'outils numériques pour la simulation de dispositifs avancés à base de matériaux alternatifs au Si : l’InGaAs et le SiGe. C'est un travail de collaboration entre l'industrie (STMicroelectronics à Crolles) et des instituts de recherche (le CEA à Grenoble et l'IEMN à Lille). La modélisation de dispositifs MOSFET avancés pour des applications de basse puissance est étudiée, grâce à des outils prédictifs, mais efficaces et peu coûteux numériquement, qui peuvent être compatibles avec un environnement industriel. L’étude porte sur différents aspects, tels que i) les propriétés électroniques des matériaux massifs et des nanostructures, avec des outils allant de la méthode des liaisons fortes et des pseudo-potentiels empiriques, à la masse effective ; ii) les propriétés électrostatiques des capacités III-V ; iii) les propriétés de transport (mobilité effective à faible champ et vitesse de saturation) dans les films minces et les nanofils ; iv) la simulation de dispositifs conventionnels planaires FDSOI 14nm en régime linéaire et saturé. Ce travail fait usage d'une large variété d'approches et de modèles différents. Des outils basés sur une approche physique sont développés, permettant d'améliorer la capacité prédictive des modèles TCAD conventionnels, pour la modélisation des dispositifs nanoscopiques à courte longueur de grille et à base de matériaux SiGe ou InGaAs. / This PhD work aims at contributing to the development of numerical tools for advanced device simulation including alternative materials (InGaAs and SiGe). It is a collaboration work, between the industry (STMicroelectronics--Crolles) and research institutes (CEA--Grenoble and IEMN--Lille). The modeling of advanced low-power MOSFET devices is investigated with predictive, but efficient tools, that can be compatibles with an industrial TCAD framework. The study includes different aspects, such as: i) the electronic properties of bulk materials and nanostructures, with tools ranging from atomistic tight-binding and empirical pseudo-potential to effective mass model; ii) the electrostatic properties of III-V Ultra-Thin Body and bulk MOSCAPs; iii) the transport properties (low-field effective mobility and saturation velocity) of thin films and nanowires; iv) the simulation of template 14nm FDSOI devices in linear and saturation regime. This work makes use of a broad variety of approaches, models and techniques. Physical-based tools are developed, allowing to improve the predictive power of TCAD models for advanced devices with short-channel length and alternative channel materials.

Vertical-cavity surface-emitting lasers (VCSELs) areattractive as potential inexpensive high-performance emittersfor fibre-optical communication systems. Their surface-normalemission together with the small dimensions are beneficial forlow-cost fabrication since it allows on-wafer testing,simplified packaging and effective fibre-coupling. Forhigh-speed data transmission up to hundreds of metres, 850-nmVCSELs are today the technology of choice. For higher bandwidthand longer distance networks, emission at long-wavelength(1.3-1.55 µm) is required. Long-wavelength VCSELs are,however, not available since no materials system offershigh-index-contrast distributed Bragg reflectors (DBRs) as wellas high-gain active regions at such wavelengths.High-performance DBRs may be built up from AlGaAs/GaAsmultilayers, but long wavelength quantum wells (QWs) are onlywell established in the InP system. Therefore, the bestperforming devices have relied on wafer-fusion betweenInP-based QWs and AlGaAs-DBRs. More recently, however, the mainefforts have been shifted towards all-epitaxial GaAs-baseddevices, employing 1.3-µm GaInNAs QWs. In this thesis, different processing technologies forlong-wavelength VCSELs are described. This includes a thoroughinvestigation of wafer-fusion between InP and GaAs regardingelectro-optical as well as metallurgical properties, and thedevelopment of a stable low-pressure process for the selectiveoxidation of AlAs. Optimised AlGaAs/GaAs DBRs were designed andfabricated. An important and striking observation from thatstudy is that n-type doping potentially is much moredetrimental to device performance than previously anticipated.These investigations were exploited in the realisation of twonew VCSEL designs. Near-room-temperature continuous-waveoperation of a single-fused 1.55-µm VCSEL was obtained.This demonstrated the potential of InGaAsP/InP DBRs inhigh-performance VCSELs, but also revealed a high sensitivityto self-heating. Further efforts were therefore directedtowards all-epitaxial GaAs-based structures. This resulted in ahigh-performance 1215-nm VCSEL with a highly strained InGaAssingle QW. This can be viewed as a basis for longer-wavelengthVCSELs, i.e., with an emission wavelength approaching 1300 nm,either by an extensive device detuning or with GaInNAs QWs. <b>Keywords</b>: VCSEL, vertical cavity laser, semiconductorlaser, long-wavelength, DBR, oxidation, wafer fusion, InGaAs,semiconductor processing

Vertical cavity lasers (VCLs) are of great interest aslow-cost, high-performance light sources for fiber-opticcommunication systems. They have a number of advantages overconventional edge-emitting lasers, including low powerconsumption, efficient fiber coupling and wafer scalemanufacturing/testing. For high-speed data transmission overdistances up to a few hundred meters, VCLs (or arrays of VCLs)operating at 850 nm wavelength is today the technology ofchoice. While multimode fibers are successfully used in theseapplications, higher transmission bandwidth and longerdistances require single-mode fibres and longer wavelengths(1.3-1.55 µm). However, long-wavelength VCLs are as yetnot commercially available since no traditional materialssystem offers the required combination of bothhigh-index-contrast distributed Bragg reflectors (DBRs) andhigh-gain active regions. Earlier work on long-wavelength VCLshas therefore focused on hybrid techniques, such as waferfusion between InP-based QWs and AlGaAs DBRs, but more recentlythe main interest in this field has shifted towardsall-epitaxial GaAs-based devices employing novel 1.3-µmactive materials. Among these, strained GaInNAs/GaAs QWs aregenerally considered one of the most promising approaches andhave received a great deal of interest. The aim of this thesis is to investigate monolithicGaAs-based long-wavelength (&gt;1.2 µm) VCLs with InGaAsor GaInNAs QW active regions. Laser structures - or partsthereof - have been grown by metal-organic vapor phase epitaxy(MOVPE) and characterized by various techniques, such ashigh-resolution x-ray diffraction (XRD), photoluminescence(PL), atomic force microscopy, and secondary ion massspectroscopy (SIMS). High accuracy reflectance measurementsrevealed that n-type doping is much more detrimental to theperformance of AlGaAs DBRs than previously anticipated. Asystematic investigation was also made of the deleteriouseffects of buried Al-containing layers, such as AlGaAs DBRs, onthe optical and structural properties of subsequently grownGaInNAs QWs. Both these problems, with their potential bearingon VCL fabrication, are reduced by lowering the DBR growthtemperature. Record-long emission wavelength InGaAs VCLs were fabricatedusing an extensive gain-cavity detuning. The cavity resonancecondition just below 1270 nm wavelength occurs at the farlong-wavelength side of the gain curve. Still, the gain is highenough to yield threshold currents in the low mA-regime and amaximum output power exceeding 1 mW, depending on devicediameter. Direct modulation experiments were performed on1260-nm devices at 10 Gb/s in a back-to-back configuration withopen, symmetric eye diagrams, indicating their potential foruse in high-speed transmission applications. These devices arein compliance with the wavelength requirements of emerging10-Gb/s Ethernet and SONET OC-192 standards and may turn out tobe a viable alternative to GaInNAs VCLs. <b>Keywords:</b>GaInNAs, InGaAs, quantum wells, MOVPE, MOCVD,vertical cavity laser, VCSEL, long-wavelength, epitaxy, XRD,DBR

The purpose of our research is growing the coupled InGaAs quantum dots on the n-type substrate by molecular beam epitaxy in laboratory, and we choose 5,10 and 15 nanometers to be the thicknesses of GaAs spacer between the quantum dots layer. Due to the couple effect, we hope to realize the theorem of intermediate band proved by Luque and Marti. We measure the characteristic of samples by electroluminescence spectra, photoelectric current spectra, electrical absorption spectra and electro reflectance spectra in laboratory; moreover, we acquire the basic parameters of solar cell by AM1.5G for analyzing. From the basic parameters of solar cell, we know that the quantum dots can enhance the photocurrent by absorbing additional photons , however, the strain caused by quantum dots would decay the open voltage seriously, so that the efficiency always under the baseline. Each efficiency of 9-stack QDs are 4.3%(c494),5.1%(c519),5.3% (c520),and each efficiency of 9-stack Dwells are 3.9%(c524),4.2%(c525),4.7%(c526), and 10-stack QDs(5nm) is 2.9%(c514),and 12-stack QDs(10nm) is 4.48%(c538),and 12-stack QDs(15nm) is 5.89%. The break through of this paper is that the efficiency of c529¡]VOC=0.64V,JSC=11.97mA/cm2,FF=67%,£b=5.89%¡^is higher than GaAs¡]VOC =0.87 V, JSC =7.4 mA/cm2,FF=72.3%,£b=5.6%¡^,and we attribute this performance to its good quality of miniband, because the current can be enhanced a lot, and it will make up for the lose of open voltage and filling factor, so that the efficiency can be higher than GaAs baseline.

The purpose of the thesis is enhancing efficiency of asymmetric quantum dots (AMQD) solar cells. The AMQD structures are grown on the n-type GaAs substrate by (MBE). In order to enhance the photovoltaic characteristics, we introduce InGaAs quantum well (QW) and modulation doping in the well to investigate effect of the strain relief and built-in electric field in the active layer. In our experiment, we analyze the optical property of AMQD structures by photoluminescence measurement system, and then decompose emission wavelength by Gaussian fitting to find optical characteristics of each single layer quantum dots. Besides, we also measure photocurrent spectra, external quantum efficiency, electrical absorption, and electro reflectance spectra to discuss carrier transition inside AMQD structure . Finally, we acquire the photovoltaic basic parameter under one sun. The results show that QDs provide additional photocurrent via absorbing extra photons, but the open circuit voltage decrease seriously due to the accumulated strains. So as to relieve the strains and enhance carriers extraction, we introduce QW layers with different growth temperatures and change the modulation doping concentrations . From the results, the higher growth temperature for QW diminishes accumulated strains, and the higher p-type modulation doping concentration indicates an extraction enhancement due to the stronger built-in electric field. By optimizing QW growth conditions, the efficiency has overtaken GaAs baseline cells. In addition, we improve the photon-excited current collection by using matrix pattern and wet etching on the device surface, the best photovoltaic characteristic shows V OC = 0.74 V, J SC = 18.82 mA/cm2, FF = 0.78, £b= 10.86%.

<p>Vertical-cavity surface-emitting lasers (VCSELs) areattractive as potential inexpensive high-performance emittersfor fibre-optical communication systems. Their surface-normalemission together with the small dimensions are beneficial forlow-cost fabrication since it allows on-wafer testing,simplified packaging and effective fibre-coupling. Forhigh-speed data transmission up to hundreds of metres, 850-nmVCSELs are today the technology of choice. For higher bandwidthand longer distance networks, emission at long-wavelength(1.3-1.55 µm) is required. Long-wavelength VCSELs are,however, not available since no materials system offershigh-index-contrast distributed Bragg reflectors (DBRs) as wellas high-gain active regions at such wavelengths.High-performance DBRs may be built up from AlGaAs/GaAsmultilayers, but long wavelength quantum wells (QWs) are onlywell established in the InP system. Therefore, the bestperforming devices have relied on wafer-fusion betweenInP-based QWs and AlGaAs-DBRs. More recently, however, the mainefforts have been shifted towards all-epitaxial GaAs-baseddevices, employing 1.3-µm GaInNAs QWs.</p><p>In this thesis, different processing technologies forlong-wavelength VCSELs are described. This includes a thoroughinvestigation of wafer-fusion between InP and GaAs regardingelectro-optical as well as metallurgical properties, and thedevelopment of a stable low-pressure process for the selectiveoxidation of AlAs. Optimised AlGaAs/GaAs DBRs were designed andfabricated. An important and striking observation from thatstudy is that n-type doping potentially is much moredetrimental to device performance than previously anticipated.These investigations were exploited in the realisation of twonew VCSEL designs. Near-room-temperature continuous-waveoperation of a single-fused 1.55-µm VCSEL was obtained.This demonstrated the potential of InGaAsP/InP DBRs inhigh-performance VCSELs, but also revealed a high sensitivityto self-heating. Further efforts were therefore directedtowards all-epitaxial GaAs-based structures. This resulted in ahigh-performance 1215-nm VCSEL with a highly strained InGaAssingle QW. This can be viewed as a basis for longer-wavelengthVCSELs, i.e., with an emission wavelength approaching 1300 nm,either by an extensive device detuning or with GaInNAs QWs.</p><p><b>Keywords</b>: VCSEL, vertical cavity laser, semiconductorlaser, long-wavelength, DBR, oxidation, wafer fusion, InGaAs,semiconductor processing</p>

<p>Vertical cavity lasers (VCLs) are of great interest aslow-cost, high-performance light sources for fiber-opticcommunication systems. They have a number of advantages overconventional edge-emitting lasers, including low powerconsumption, efficient fiber coupling and wafer scalemanufacturing/testing. For high-speed data transmission overdistances up to a few hundred meters, VCLs (or arrays of VCLs)operating at 850 nm wavelength is today the technology ofchoice. While multimode fibers are successfully used in theseapplications, higher transmission bandwidth and longerdistances require single-mode fibres and longer wavelengths(1.3-1.55 µm). However, long-wavelength VCLs are as yetnot commercially available since no traditional materialssystem offers the required combination of bothhigh-index-contrast distributed Bragg reflectors (DBRs) andhigh-gain active regions. Earlier work on long-wavelength VCLshas therefore focused on hybrid techniques, such as waferfusion between InP-based QWs and AlGaAs DBRs, but more recentlythe main interest in this field has shifted towardsall-epitaxial GaAs-based devices employing novel 1.3-µmactive materials. Among these, strained GaInNAs/GaAs QWs aregenerally considered one of the most promising approaches andhave received a great deal of interest.</p><p>The aim of this thesis is to investigate monolithicGaAs-based long-wavelength (>1.2 µm) VCLs with InGaAsor GaInNAs QW active regions. Laser structures - or partsthereof - have been grown by metal-organic vapor phase epitaxy(MOVPE) and characterized by various techniques, such ashigh-resolution x-ray diffraction (XRD), photoluminescence(PL), atomic force microscopy, and secondary ion massspectroscopy (SIMS). High accuracy reflectance measurementsrevealed that n-type doping is much more detrimental to theperformance of AlGaAs DBRs than previously anticipated. Asystematic investigation was also made of the deleteriouseffects of buried Al-containing layers, such as AlGaAs DBRs, onthe optical and structural properties of subsequently grownGaInNAs QWs. Both these problems, with their potential bearingon VCL fabrication, are reduced by lowering the DBR growthtemperature.</p><p>Record-long emission wavelength InGaAs VCLs were fabricatedusing an extensive gain-cavity detuning. The cavity resonancecondition just below 1270 nm wavelength occurs at the farlong-wavelength side of the gain curve. Still, the gain is highenough to yield threshold currents in the low mA-regime and amaximum output power exceeding 1 mW, depending on devicediameter. Direct modulation experiments were performed on1260-nm devices at 10 Gb/s in a back-to-back configuration withopen, symmetric eye diagrams, indicating their potential foruse in high-speed transmission applications. These devices arein compliance with the wavelength requirements of emerging10-Gb/s Ethernet and SONET OC-192 standards and may turn out tobe a viable alternative to GaInNAs VCLs.</p><p><b>Keywords:</b>GaInNAs, InGaAs, quantum wells, MOVPE, MOCVD,vertical cavity laser, VCSEL, long-wavelength, epitaxy, XRD,DBR</p>

Quantum networks lie at the heart of distributed quantum computing and secure quantum communication - research areas that have seen a strong increase of interest over the last decade. Their basic architecture consist of stationary nodes composed of quantum processors which are linked via photonic channels. The key requirement, and at the same time the most demanding challenge, is the efficient distribution of entanglement between distant nodes. The two ground states of single spins confined in self-assembled InGaAs quantum dots provide an effective two-level system for the implementation of quantum bits. Moreover, they offer strong transition dipole moments with outstanding photonic properties allowing for the realisation of close to ideal, high-bandwidth spin-photon interfaces. These properties are combined with the benefits of working in the solid state, such as scalability and integrability of devices, to form a promising candidate for the implementation of fast entanglement distribution. In this dissertation we provide the first implementation of a unit cell of a quantum network based on single electron spins in InGaAs. We use a probabilistic scheme based on spin-photon entanglement and the erasure of which path information to project the two distant spins into a maximally entangled Bell state. The successful generation of entanglement is verified through a reconstruction of the final two-spin state and we achieve an average fidelity of $61.6\pm2.3\%$ at a record-high generation rate of $5.8\,\mathrm{kHz}$. One of the main constraints to the achieved fidelity is the limited coherence of the electron spin. We show that it can be extended by three orders of magnitude through decoupling techniques and develop a new measurement technique, allowing us to investigate the origins of the decoherence which has previously been obscured by nuclear feedback processes. Our results evidence that further extension of coherence is ultimately limited by intrinsic mechanisms closely related to local strain due to the growth method of self-assembled quantum dots. After establishing the intrinsic limits to the electron coherence we investigate the coherence properties of the single hole spin as an alternative two-level system with the potential for higher coherence times. We show that the hole spin coherence is indeed superior to the one of the electron and realise the first successful dynamic decoupling scheme implemented in these systems. We find that the decoherence at low external magnetic fields is still governed by coupling to the nuclear spins whereas it is dominated by electrical noise for fields exceeding a few Tesla. This noise source is extrinsic to the quantum dots and a better understanding offers the potential for further improvement of the coherence time. The findings of this work present a complete study of the coherence of the charge carriers in self-assembled quantum dots and provide the knowledge needed to improve the implementation of a quantum-dot based quantum network. In particular, the combination of spin-spin entanglement and the hole coherence times enable further research towards multidimensional photonic cluster states.

III-V semiconductors have higher charge carrier mobilities than silicon and are used in photovoltaic devices, optical sensors, and emitters. The high injection velocities obtained with III-V channels allow for faster transistors with low power consumption. However, the large-scale implementation in electronic devices is currently limited by the defective interface formed between III-Vs and their oxides. Clean III-V surfaces are highly reactive in air and form amorphous oxides that lead to a high density of dangling bonds. Satisfying these dangling bonds has been associated with an improvement in electrical performance, directing the development of strategies that decrease the surface reactivity (chemical passivation) and the density of surface states that cause power dissipation (electrical passivation). Sulfur bonds easily to III-V surfaces and has been used to chemically and electrically passivate GaAs. In this work, we investigated liquid phase sulfur chemistries in the chemical passivation of clean InSb(100) and In0.53Ga0.47As(100) surfaces terminated by their group V elements. Our strategy consisted of maximizing the number of bonds between sulfur and antimony or arsenic. A long alkane chain thiol, 1-eicosanethiol (ET, 20 carbon atoms), was used to produce a hydrophobic surface and deposit a dense organic layer by taking advantage of the van der Waals interactions between thiol molecules. The first part of the study involved the optimization of the thiol deposition process on InSb. Self-assembled alkanethiol monolayers were formed by immersing clean InSb substrates in ET solutions in ethanol for 20 h. The layers prevented the formation of detectable oxides for 20 min based on the O Auger x-ray photoelectron spectroscopy (XPS) peak. The thiol layer was completely removed by heating the surface to 227 C in vacuum. In the second part of the study, a 20 h ET deposition was performed on In0.53Ga0.47As(100), and re-oxidation was prevented for up to 4 min based on the O 1s XPS peak. The alkanethiolate layer was removed by heating the samples to 350 C in vacuum. The sulfur coverage after 20 min and 20 h ET depositions was increased by performing a second immersion in (NH4)2S without modifying the thickness of the layer. The best process studied consisted of a 20 h immersion in ET solution followed by a short (NH4)2S step, preventing the formation of oxides for up to 9 min. This is due to the presence of available surface sites and weakly bonded molecules in the layer after a long 20 h ET process. The chemical passivation effect is not uniquely influenced by surface termination, roughness, or lattice constant, but is rather a result of a combination of these factors. Future work will involve the fabrication and electrical characterization of III-V devices modified with various chemical passivation strategies.