Submitted by Regina Correa (rehecorrea@gmail.com) on 2016-09-21T18:57:01Z No. of bitstreams: 1 TeseLKSH.pdf: 62711757 bytes, checksum: dd2b481497af4493346131729910c613 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-09-23T18:36:35Z (GMT) No. of bitstreams: 1 TeseLKSH.pdf: 62711757 bytes, checksum: dd2b481497af4493346131729910c613 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-09-23T18:36:40Z (GMT) No. of bitstreams: 1 TeseLKSH.pdf: 62711757 bytes, checksum: dd2b481497af4493346131729910c613 (MD5) / Made available in DSpace on 2016-09-23T18:36:47Z (GMT). No. of bitstreams: 1 TeseLKSH.pdf: 62711757 bytes, checksum: dd2b481497af4493346131729910c613 (MD5) Previous issue date: 2015-12-14 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Defects play a fundamental issue on physical properties of wide bandgap semiconductors (WBS) due to the possibility of application in high temperature. This work shows the features of the defects in di erent kind of WBS grown/synthesized by solid state reaction, Pechini method and Molecular Beam Epitaxy. First we investigated the structural and optical properties of erbium- and manganese-doped strontium aluminates (SrAl2O4). The fundamental optical transitions due to Er3+ and Mn2+ are typical features of the well-diluted doping process. A signi cant enhancement of the Er3+ optical emission band at 1530 nm was observed when the matrix is co-doped with Mn. A model of energy transfer mechanism from Mn2+ to Er3+ is proposed to explain the experimental results. We present also the e ect of preparation and annealing conditions on the properties of Nb2O5. The increase of oxygen vacancies causes an inrreversible phase transition from pseudohexagonal to orthorhombic phase, and they are responsible for the increase in the e ective magnetic moments related to paramagnetic behavior. Finally, we explore the e ects of the interfaces imperfections of quantum wells of cubic GaN alloys. An indication of localized states in low temperature is observed. A decrease of this e ects happens when we increase the number of interfaces. Furthermore, magneto-PL measurements show a higher spin polarizations in the donor-acceptor impurities in the bulk of c-GaN corroborate for the importance of the defects. This work shows that the defects engineering in WBS are fundametal for developing new technologies in spintronics and optoelectronic eld. / Os defeitos em semicondutores acarretam mudanças nas propriedades físicas do material, tendo uma grande importância nos semicondutores que possuem alta energia de gap devido à possibilidade de aplicação em alta temperatura. No presente trabalho, estudamos as características atribuídas à presença de defeitos em diferentes semicondutores de gap largo, crescidos/sintetizados por reação de estado sólido, método de Pechini e epitaxia de feixes moleculares. Primeiramente, investigamos as propriedades ópticas e estruturais do aluminato de estrôncio (SrAl2O4) dopado com érbio e co-dopado com manganês. Neste caso, um aumento na emissão do Er3+ na região do infravermelho (1530 nm) foi observado na matriz com co-dopagem de manganês. Atribuímos este efeito ao mecanismo de transferência de energia do íon de Mn2+ para o Er3+. Também estudamos os efeitos das condições de preparação e tratamento térmico nas propriedades de pentóxido de nióbio (Nb2O5). Nossa investigação, mostra que vacâncias de oxigênio causam transições irreversíveis da fase pseudo-hexagonal para a ortorrômbica. Além disso, esses defeitos são responsáveis por um acréscimo de momentos magnéticos efetivos, relacionados com o comportamento paramagnético. Por último, averiguamos os efeitos de imperfeições entre as interfaces de poços quânticos de ligas de nitreto de gálio cúbico (cGaN). Esse estudo indicou uma diminuição de estados localizados, em baixa temperatura, com a mudança na quantidade de interfaces do QW. No caso do c GaN bulk, foi observado uma maior polarização de spin na emissão óptica da impureza doadora-aceitadora. Este trabalho evidencia a importância da engenharia de defeitos em semicondutores de gap largo para o aperfeiçoamento de tecnologias ligadas à área de optoeletrônica e spintrônica.

Orientadores: Evandro Conforti, Napoleão dos Santos Ribeiro / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-16T11:21:52Z (GMT). No. of bitstreams: 1 Figueiredo_RafaelCarvalho_M.pdf: 10424465 bytes, checksum: b93608291545cc7a53e52e2050a3dc39 (MD5) Previous issue date: 2010 / Resumo: Apresenta-se a modelagem de um circuito elétrico equivalente e a extração de parâmetros de amplificadores ópticos a semicondutor (SOA), a partir de um modelo para lasers semicondutores. Foi realizado um estudo do comportamento da impedância de um SOA em chip, sem encapsulamento, em função da corrente de polarização e em ampla faixa de frequência . de 300 kHz a 40 GHz. A modelagem do circuito equivalente da montagem, a qual é cascateada com os modelos da região ativa do SOA, é apresentada para correntes abaixo e acima da operação em transparência. A metodologia utilizada para a extração dos parâmetros dos elementos parasitas que compõe o circuito é descrita; resultados obtidos através de simulações em programa comercial (Agilent ADS) são comparados com medidas experimentais obtidas em mesa óptica. São apresentados ainda resultados teóricos da impedância do SOA quando desconsiderada a presença dos elementos parasitas da montagem. A modelagem e extração dos parâmetros realizada para o chip foi repetida para SOAs encapsulados, também apresentando boa concordância entre teoria e experimento, reforçando a viabilidade da abordagem utilizada / Abstract: The equivalent electric circuits and its parameters.extraction of semiconductor optical amplifiers (SOA) are attained based on a diode-laser model. Additionally, the impedance behavior of a SOA-chip (without package) was measured as function of the bias current in wide frequency range, from 300 kHz to 40 GHz. In these procedures, the microwave setup used for the SOA current injection was also characterized and its equivalent circuit obtained. Next, a theoretical analysis is developed for this setup for currents below and above the transparency condition. A methodology for the parameters extraction of parasitic elements is also described, as well as the results obtained through simulations using the Agilent ADS software, compared with the experimental data. The optical bench used in the experiments is also described, and theoretical results illustrates the SOA impedance without parasitic elements. The equivalent circuits with parameters.extraction were also obtained for packaged SOAs, with good agreement between theory and experiment, conforming the employed methodology / Mestrado / Telecomunicações e Telemática / Mestre em Engenharia Elétrica

Visando atender a massificação das tecnologias da informação e comunicação (TIC) por meio de um aproveitamento mais eficiente da infra-estrutura de fibras ópticas, as redes ópticas DWDM vem passando por significativa evolução de capacidade, com base no uso de formatos de modulação avançados, para canais operando em taxas de 100 Gb/s e superiores, bem como no emprego de topologias dinâmicas e reconfiguráveis. Estas redes ópticas de nova geração impõe novos requisitos de desempenho aos amplificadores ópticos. Especificamente, as características dinâmicas da rede tornam obrigatório o uso de esquemas de controle que assegurem estrita planicidade espectral de ganho enquanto o emprego de formatos de modulação avançados e de alta ordem requer margens mais estreitas em termos da relação sinal-ruído aceitável para detecção dos sinais recebidos. Neste contexto, esta tese propõe e avalia experimentalmente uma topologia de amplificação óptica híbrida Raman/EDFA, introduzindo um novo esquema de controle automático de ganho e apresentando desempenho superior aos amplificadores atualmente usados em redes DWDM reconfiguráveis. O amplificador óptico híbrido desenvolvido baseia-se em um estágio Raman distribuído contra-propagante, com excelente desempenho de figura de ruído (porém com baixa eficiência de conversão de bombeio em amplificação - PCE) seguido de um estágio EDFA, que assegura alta potência de saída, devido a sua elevada PCE. Ganho espectral plano foi obtido por meio de uma técnica de controle automático de ganho inovadora, baseada na atuação paralela e independente de duas malhas de controle automático de ganho, uma primeira aplicada ao estágio de amplificação Raman visando ganho-alvo variável com baixa variação espectral, enquanto outra malha de controle de ganho visa fornecer ganho alvo fixo ao estágio EDFA, com alta potência de saída. / Seeking to support the massive deployment of information and communication technologies (ICTs) by means of a more efficient usage of the optical fiber infrastructure, DWDM optical networks have been undergoing a significant capacity evolution, by using advanced modulation formats for optical channels operating at data rates of 100 Gb/s and beyond, as well as by employing dynamic and reconfigurable network topologies. These new generation optical networks impose new performance benchmarks on the optical amplifiers. Specifically, the dynamic characteristics of the network make mandatory the deployment of control schemes which assure stringent optical gain spectral flatness while the usage of high-order advanced modulation formats translate into more strict margins of signal-to-noise ratios for the detected signals. In this context, this thesis proposes and experimentally evaluates an hybrid Raman/ EDFA optical amplifier topology, introducing a novel automatic gain control scheme and demonstrating improved performance over the optical amplifiers already in use in DWDM reconfigurable networks. The developed hybrid optical amplifier is based on a distributed counter-propagating Raman stage, displaying excellent noise figure performance (albeit presenting low conversion efficiency - PCE) followed by an EDFA stage, which assures high output power, due to its high PCE. Flat spectral gain was achieved by means of a novel gain control technique, based on the parallel and independently acting of two control schemes, the first applied over the Raman amplifying stage, aiming at a variable target gain and low spectral gain ripple, while the other seeks to attain a fixed target gain at the EDFA, assuring a high output power.

Made available in DSpace on 2016-03-15T19:37:49Z (GMT). No. of bitstreams: 1 Lucia Akemi Saito -EE2006.pdf: 2213180 bytes, checksum: d5a6d162d17171e2b4cf99c0fa4797ce (MD5) Previous issue date: 2006-08-02 / Conselho Nacional de Desenvolvimento Científico e Tecnológico / Most of researches about Raman amplifier had been made in C and L bands (1530-1625 nm), which is possible to found more data of Raman efficiency gain and the requirements for design is consolidated. The first issue that should be considered when the amplifier is to use in another band, like our project (O-band) is to consider the decrease of effective area and wavelength in the calculation of Raman efficiency, not only the pump wavelength. We have two configuration types: for the first one, the gain is obtained in the transmission fiber and for the second that is demonstrated in this work, we need special fibers as DSF, DCF and Raman fiber to amplifier the signals. Note that the amplification can be obtained in any band of optical spectrum, which depends on the pump wavelength that is used for made it. Our studies demonstrated that the discrete Raman amplifier that was made of Raman fiber of OFS Fitel Denmark was more efficient in short wavelengths and has more improvement when it was working in O-band. In spite of the higher attenuation, we can have higher Raman gain efficiency (CR) because the effective area (Aeff) decreased in short wavelengths. The characteristics of Raman fiber were studied when we found the Raman gain efficiency peak of 3.9 (W.km)-1 for a pump laser of 1240 nm wavelength. For this case, the gain of O-band amplifier was about 50 % higher when we compared with the C-band. For a CWDM system, we need to design a multi-pump amplifier. It is necessary four pump lasers to amplifier a bandwidth of 70 nm and six pump lasers to cover all O-band. The gain value depends on the pump power and if we want a good result, we should verify the total power of channels in the input of amplifier to avoid the device saturation. In addition, the noise figure of Raman amplifier was studied and then we found results that demonstrated higher noise values in short wavelengths specially, when we need to locate some pump lasers between signal wavelengths. / A maior parte das pesquisas realizadas sobre amplificadores Raman tem seus estudos centrados nas bandas C (1530 a 1565 nm) e L (1565 a 1625 nm), regiões onde os dados de eficiência de ganho Raman e a parametrização dos amplificadores têm seus estudos consolidados. Um ponto importante para ser considerado quando se deseja trabalhar numa banda diferente, como no caso a banda O (1260 a 1360 nm), é que a alteração não é apenas no comprimento de onda do laser de bombeio. Deve-se considerar o decréscimo da área efetiva e do comprimento de onda que influenciará diretamente no valor da eficiência e conseqüentemente no próprio ganho do amplificador Raman. Alguns amplificadores Raman utilizam a própria fibra de transmissão e outros têm como meio de amplificação fibras especiais como DSF, DCF ou a própria fibra Raman como é demonstrado neste trabalho. Sabe-se, no entanto, que a amplificação Raman pode ocorrer em qualquer banda do espectro óptico sendo necessário para a sua montagem, apenas alterar o comprimento de onda do laser de geração do efeito. Este trabalho demonstrou que o Amplificador Raman Discreto utilizando a fibra Raman da OFS Fitel Denmark é mais eficiente em comprimentos de onda menores e com melhor aproveitamento na região compreendida pela banda O . Nossos estudos demonstraram que apesar do aumento da atenuação nesta região, pode-se obter maior Eficiência de Ganho Raman (CR) devido ao decréscimo da Área Efetiva (Aeff) em comprimentos de onda curtos. A variação das características da fibra Raman para a banda O foram analisadas sendo encontrado um pico de Eficiência de Ganho Raman de 3,9 (W.km)-1 para um laser de bombeio de 1240 nm resultando em torno de 50 % a mais de Ganho no amplificador quando comparado com a banda C . Analisando o amplificador utilizando múltiplos lasers de bombeio, verificou-se que para uma aplicação em sistema CWDM seriam necessários 4 lasers para amplificação numa faixa de 70 nm e 6 lasers para cobrir toda a banda O sendo que a magnitude de amplificação depende diretamente da potência destes lasers devendo-se observar a potência total dos canais na entrada para evitar a saturação do dispositivo. Foi analisada a figura de ruído do amplificador Raman sendo encontrados valores mais altos em comprimentos de onda curtos e de maior magnitude quando há lasers de bombeio entre os comprimentos de onda de sinal.

Over the last decade, rapid growth of broadband services necessitated research aimed at increasing transmission capacity in fiber-optic communication systems. Wavelength division multiplexing (WDM) technology has been widely used in fiber-optic systems to fully utilize fiber transmission bandwidth. Among optical amplifiers for WDM transmission, semiconductor optical amplifier (SOA) is a promising candidate, thanks to its broad bandwidth, compact size, and low cost. In transmission systems using SOAs, due to their large noise figures, high signal launching powers are required to ensure reasonable optical signal-to-noise ratio of the received signals. Hence the SOAs are operated in the saturation region and the signals will suffer from SOA impairments including self-gain modulation, self-phase modulation, and inter channel crosstalk effects such as cross-gain modulation, cross-phase modulation, and four-wave mixing in WDM. One possibility to circumvent these nonlinear impairments is to use constant-intensity modulation format in the 1310 nm window where dispersion is also negligible. In this dissertation, differential phase-shift keying (DPSK) WDM transmission in the 1310 nm window using SOAs was first considered to increase the capacity of existing telecommunication network. A WDM transmission of 4 x 10 Gbit/s DPSK signals over 540 km standard single mode fiber (SSMF) using cascaded SOAs was demonstrated in a recirculating loop. In order to increase the transmission reach of such WDM systems, those SOA impairments must be compensated. To do so, an accurate model for quantum-dot (QD) SOA must be established. In this dissertation, the QD-SOA was modeled with the assumption of overall charge neutrality. Static gain was calculated. Optical modulation response and nonlinear phase noise were studied semi-analytically based on small-signal analysis. The quantitative studies show that an ultrafast gain recovery time of ~0.1 ps can be achieved when QD-SOAs are under high current injection, which leads to high saturation output power. However more nonlinear phase noise is induced when the QD-SOAs are used in the transmission systems operating at 10 Gbit/s or 40 Gbit/s. Electronic post-compensation for SOA impairments using coherent detection and digital signal processing (DSP) was investigated next in this dissertation. An on-off keying transmission over 100 km SSMF using three SOAs at 1.3 [micrometer] were demonstrated experimentally with direct detection and SOA impairment compensation. The data pattern effect of the signal was compensated effectively. Both optimum launching power and Q-factor were improved by 8 dB. For advanced modulation formats involving phase modulation or in transmission windows with large dispersion, coherent detection must be used and fiber impairments in WDM systems need to be compensated as well. The proposed fiber impairment compensation is based on digital backward propagation. The corresponding DSP implementation was described and the required calculations as well as system latency were derived. Finally joint SOA and fiber impairment compensations were experimentally demonstrated for an amplitude-phase-shift keying transmission.

Os sistemas ópticos atuais, baseados em fibras monomodo, operam próximos ao limite teórico da capacidade. Sistemas ópticos baseados em multiplexação modal (Mode Division Multiplexing – MDM) possibilitam o aumento da capacidade do sistema por meio do uso de fibras de poucos modos. Nestes sistemas, a propriedade de ortogonalidade entre os modos propagantes permite que cada modo espacial carregue um sinal óptico específico. O amplificador à fibra dopada com érbio (Erbium-Doped Fiber Amplifier – EDFA) segue fundamental para assegurar a transmissão em longas distâncias. No entanto, devido às distintas distribuições de intensidade dos modos que compõem o sinal de entrada, cada modo experimenta diferentes valores de ganho. Desta forma, o objetivo principal no projeto de EDFAs de poucos modos (Few-Mode Erbium-Doped Fiber Amplifier – FM-EDFA) é determinar os melhores parâmetros opto-geométricos da fibra para produzir uma amplificação eficiente. A metodologia normalmente empregada é baseada na resolução das equações de taxa e de propagação. Nesta tese, é proposta uma metodologia alternativa de projeto de FM-EDFA, baseada em uma nova figura de mérito. Este parâmetro quantifica o nível de inversão da população dos íons na fibra a partir da integral de superposição (overlap integral), considerando tanto o perfil de dopagem da fibra dopada com érbio para poucos modos (Few-Mode Erbium-Doped Fiber – FM-EDF) quanto as distribuições de intensidade dos sinais de entrada e de bombeio. A aplicação desta metodologia permite reduzir, em cerca de 25-40 vezes, o número de resoluções das equações de taxa e de propagação e, consequentemente, diminuir o tempo de processamento e reduzir o esforço computacional. Como consequência da maior velocidade de processamento, torna-se possível a aplicação de métodos de otimização mais rigorosos, permitindo uma busca em um espaço irrestrito de soluções. Especificamente, a partir de uma metodologia baseada em algoritmos genéticos, obteve-se um perfil de dopagem otimizado. É também demonstrado que os perfis com geometria circular exibem melhores características, como excelente desempenho do FM-EDFA e maior facilidade de fabricação. Por meio da análise da figura de mérito, é mostrado que o desempenho do FM-EDFA é afetado pelas características do modo de bombeio. Finalmente, o desempenho de um sistema óptico MDM é avaliado, simulado por meio da integração entre as ferramentas MatLab® e VPItransmissionMakerTM, comprovando a necessidade do projeto de um amplificador específico para sistemas MDM. / Modern optical systems based on single-mode fiber, operate close to the theoretical capacity limit. By using few-mode fibers, optical systems based on modal division multiplexing (MDM) allows increased system capacity. In these systems, orthogonality between the propagating modes allows each spatial mode to carry a specific optical signal. The erbium doped fiber amplifier (EDFA) remains essential to ensure long distance transmission. However, due to the distinct intensity profile distributions of the modes which comprise the input signal, each mode experiences a different value of optical gain. Thus, the main objective in the few-mode EDFA design (FM-EDFA) is to determine the best opto-geometrical fiber parameters in order to produce an efficient amplification. The methodology normally used is based on the simultaneous resolution of the rate and propagation equations. In this thesis, we propose an alternative methodology for the FM-EDFA design, based on a new figure of merit which quantifies the level of population inversion for the Er ions in the fiber, by means of a overlap integral considering both the doping profile of the few-mode erbium doped fiber (FM-EDF) as well as the intensity distributions of the optical signals and pump beams. This methodology reduces, by a factor of 25-40, the number of resolutions of the rate and propagation equations, thereby decreasing processing time and computational effort. As a consequence of the improved processing speed, it becomes possible to apply more rigorous optimization methods in an unrestricted solution space. Specifically, by using a genetic algorithm technique, we obtained an optimized doping profile. It is also shown that profiles with circular geometry exhibit improved features, such as excellent FM-EDFA performance and ease of manufacturing. By analyzing the figure of merit, it is shown that the FM-EDFA performance is affected by the characteristics of the pump mode. Finally, the performance of an MDM optical system is evaluated, by integrating Matlab and VPI simulation tools, to demonstrate the need for specific amplifier designs in MDM systems.

Un grand nombre d’applications réclament des sources laser en régime impulsionnel toujours plus puissantes et énergétiques. Les progrès continus des technologies laser permettent non seulement d’améliorer les performances de l’outil laser mais aussi d’ouvrir la voie à de nouvelles applications. Cependant, l'augmentation de la puissance des sources laser est aussi accompagnée par une complexification des systèmes. Cette thèse porte sur l’étude d’amplificateurs laser de puissances qui se distinguent par la simplicité de leurs architectures : avec un ou deux passages dans le milieu laser. Dans la première partie, nous étudions le potentiel de la fibre cristalline Yb: YAG pompée par diode en tant qu’amplificateur. Les effets de confinement de l’intensité de pompe au centre de la fibre cristalline par guidage sont étudies théoriquement et expérimentalement. Deux expériences démontrent ensuite l’intérêt de la fibre cristalline Yb:YAG en tant qu’amplificateur de puissance de sources laser à fibres, l’une en régime femtoseconde et l’autre avec un laser mono-fréquence. Par ailleurs, nous explorons le potentiel de notre concept en régime de forte puissance. Une puissance de 250 W en oscillateur et une extraction de 100 W en amplificateur ont été obtenues avec une diode de pompe de 600 W. La deuxième partie traite de l’étude d’amplificateurs à base de cristaux de Nd:YVO4. Le dimensionnement de notre système est réalisé en s’appuyant sur une étude des propriétés du Nd:YVO4. L’amplificateur ainsi obtenu affiche des performances inédites qui se caractérisent par un très fort gain optique (40-60 dB) couplé à une forte extraction de puissance moyenne (10 à 15 W). / A wide range of scientific and industrial applications require pulsed laser sources delivering increasing amount of powers and pulse energies. Continuous progresses in the field of laser technology do not only bring significant process efficiency improvements but also allow developing new applications. However, the complexity of laser sources has significantly increased over the years together with their performance. In contrast, this work focuses on power amplifier architectures which are particularly simple (with one or two passes). In the first part, we study the potential of Yb:YAG single crystal fibers. Pump intensity confinement by the pump beam guiding is studied in details both theoretically and experimentally. Two experiments demonstrate the interest of Yb:YAG single crystal fibers as a power amplifier for fiber based laser sources, one in femtosecond regime and the other one with a single frequency laser. Furthermore, the potential of our architecture is also explored in high power regime. An output power of 250 W for the oscillator and 100 W power extraction with the amplifier were obtained with a 600 W pump diode. The second part describes the study of laser amplifiers using longitudinally pumped Nd:YVO4 bulk crystals. We study the effect of temperature increase in the laser crystal and evaluate the influence of the doping concentration and excited state population on the heat load. Guided by the conclusion of our study, we design a high gain amplifier characterized by very high optical gain (40-60 dB) coupled to a high average power extraction (10 to 15 W)

This dissertation includes two parallel topics: optical modeling of wavelength converters based on semiconductor optical amplifiers (SOA) and optical modeling for LEDs and solid state lighting. A steady-state numerical model of wavelength converters based on cross-gain SOAs is developed. In this model, a new model of the gain coefficient is applied. Each physical variable, such as the carrier density, gain coefficient, differential gain, and internal loss, spatially varies across the SOA cavity and is numerically calculated throughout the device. Increased accuracy over previous studies is achieved by including such spatial variations. This model predicts wavelength-dependent characteristics of a wavelength converter of the SOA in both large and small signal regimes. Some key performance factors of SOA wavelength converters. A hybrid method incorporating both guided wave optics and optical ray tracing is also developed to model LEDs and solid state lighting. This method can model either single wavelength or dual-wavelength LED structures with different die shapes and packages. The waveguide and diffraction optics are mainly used to model the near-field optics inside LED chips and its vicinity and to identify guided modes and leakage modes. Geometrical ray tracing is applied to model the far-field pattern and light interactions at different material interfaces, such as LED chip structures, LED package materials, and light scattering at those rough surfaces and textures. To improve LED light extraction efficiency, different LED die shapes and device structures can also be optimized using this method. New technologies for future research on SOAs and LEDs are also proposed.

Os sistemas ópticos atuais, baseados em fibras monomodo, operam próximos ao limite teórico da capacidade. Sistemas ópticos baseados em multiplexação modal (Mode Division Multiplexing – MDM) possibilitam o aumento da capacidade do sistema por meio do uso de fibras de poucos modos. Nestes sistemas, a propriedade de ortogonalidade entre os modos propagantes permite que cada modo espacial carregue um sinal óptico específico. O amplificador à fibra dopada com érbio (Erbium-Doped Fiber Amplifier – EDFA) segue fundamental para assegurar a transmissão em longas distâncias. No entanto, devido às distintas distribuições de intensidade dos modos que compõem o sinal de entrada, cada modo experimenta diferentes valores de ganho. Desta forma, o objetivo principal no projeto de EDFAs de poucos modos (Few-Mode Erbium-Doped Fiber Amplifier – FM-EDFA) é determinar os melhores parâmetros opto-geométricos da fibra para produzir uma amplificação eficiente. A metodologia normalmente empregada é baseada na resolução das equações de taxa e de propagação. Nesta tese, é proposta uma metodologia alternativa de projeto de FM-EDFA, baseada em uma nova figura de mérito. Este parâmetro quantifica o nível de inversão da população dos íons na fibra a partir da integral de superposição (overlap integral), considerando tanto o perfil de dopagem da fibra dopada com érbio para poucos modos (Few-Mode Erbium-Doped Fiber – FM-EDF) quanto as distribuições de intensidade dos sinais de entrada e de bombeio. A aplicação desta metodologia permite reduzir, em cerca de 25-40 vezes, o número de resoluções das equações de taxa e de propagação e, consequentemente, diminuir o tempo de processamento e reduzir o esforço computacional. Como consequência da maior velocidade de processamento, torna-se possível a aplicação de métodos de otimização mais rigorosos, permitindo uma busca em um espaço irrestrito de soluções. Especificamente, a partir de uma metodologia baseada em algoritmos genéticos, obteve-se um perfil de dopagem otimizado. É também demonstrado que os perfis com geometria circular exibem melhores características, como excelente desempenho do FM-EDFA e maior facilidade de fabricação. Por meio da análise da figura de mérito, é mostrado que o desempenho do FM-EDFA é afetado pelas características do modo de bombeio. Finalmente, o desempenho de um sistema óptico MDM é avaliado, simulado por meio da integração entre as ferramentas MatLab® e VPItransmissionMakerTM, comprovando a necessidade do projeto de um amplificador específico para sistemas MDM. / Modern optical systems based on single-mode fiber, operate close to the theoretical capacity limit. By using few-mode fibers, optical systems based on modal division multiplexing (MDM) allows increased system capacity. In these systems, orthogonality between the propagating modes allows each spatial mode to carry a specific optical signal. The erbium doped fiber amplifier (EDFA) remains essential to ensure long distance transmission. However, due to the distinct intensity profile distributions of the modes which comprise the input signal, each mode experiences a different value of optical gain. Thus, the main objective in the few-mode EDFA design (FM-EDFA) is to determine the best opto-geometrical fiber parameters in order to produce an efficient amplification. The methodology normally used is based on the simultaneous resolution of the rate and propagation equations. In this thesis, we propose an alternative methodology for the FM-EDFA design, based on a new figure of merit which quantifies the level of population inversion for the Er ions in the fiber, by means of a overlap integral considering both the doping profile of the few-mode erbium doped fiber (FM-EDF) as well as the intensity distributions of the optical signals and pump beams. This methodology reduces, by a factor of 25-40, the number of resolutions of the rate and propagation equations, thereby decreasing processing time and computational effort. As a consequence of the improved processing speed, it becomes possible to apply more rigorous optimization methods in an unrestricted solution space. Specifically, by using a genetic algorithm technique, we obtained an optimized doping profile. It is also shown that profiles with circular geometry exhibit improved features, such as excellent FM-EDFA performance and ease of manufacturing. By analyzing the figure of merit, it is shown that the FM-EDFA performance is affected by the characteristics of the pump mode. Finally, the performance of an MDM optical system is evaluated, by integrating Matlab and VPI simulation tools, to demonstrate the need for specific amplifier designs in MDM systems.

Au cours des dernières décennies, le monde a subi une révolution majeure qui a profondément affecté comment on utilise les réseaux de communication. De nouveaux services et applications ont été apparus, tels que les réseaux sociaux, les jeux en ligne ou le streaming en direct, qui exigent une augmentation constante de la capacité des systèmes optiques. La motivation de ce travail est donc d'étudier la mise en œuvre d'un nouvel amplificateur SOA à très large bande avec une bande passante de plus de 100 nm afin d’étendre la capacité du système. L'utilisation de l'amplification SOA change complètement le paradigme dans la conception d'un système optique puisque toutes les dégradations ajoutées par la SOA doivent être considérées. Ainsi, la recherche d'un modèle analytique ou numérique capable de caractériser la nature non linéaire de ce dispositif est d'abord étudiée. Ensuite, on montre comment un SOA bien conçu peut non seulement amplifier un signal à large bande, mais également surmonter certains des principaux inconvénients du SOA devant EDFA. Finalement, on évalue la capacité de ce nouveau UWB SOA pour les applications d'interconnexion de centres de données avec deux expériences en transmettant jusqu'à 115 Tbps de données dans une liaison à bande passante continue de 100 nm sur 100 km de fibre et en testant la stabilité du système avec cartes de ligne en temps réelle entre deux points de présence (POP) de Facebook déployés dans la région parisienne / Over the last few decades the world has undergone a major revolution that has deeply affected the way we use communication networks. New services and applications have appeared demanding a constant increase of the channel capacity. In this period, optical systems have been upgraded at pair with advanced signal processing techniques which have permitted the increase of the spectral efficiency approaching the system capacity to the fundamental limit. It is because is becoming extremely challenging to keep growing the system capacity by this means. In this work, an orthogonal direction is studied to further increase the fibre capacity: extending the optical bandwidth. With this purpose, the use of semiconductor optical amplifiers (SOA) is investigated to be implemented in future ultra-wideband (UWB) systems. The use of SOA amplification changes completely the paradigm in the design of an optical system since all the impairments added by the SOA must be considered. In this work, we assess the reservoir model, a simple yet powerful model, to analyze numerically the nonlinear regime of the SOA for WDM systems. We also show for the first that the linewidth enhancement factor of an SOA can be estimated with a coherent receiver. Finally, it is also studied how the correlation between channels degrades significantly the performance of the SOA and the inclusion of a decorrelation fibre is investigated. The conception of a UWB system is then studied. We characterize a novel ultra-wideband SOA developed by the French project CALIPSO which presents high gain in a 100-nm optical bandwidth with high output saturation power and 6-8 dB of noise figure. We analyze its nonlinear regime for WDM systems and we show for QPSK and 16 QAM modulation formats that the input saturation power can be overtaken by serveral dBs without important nonlinear penalty. On the other hand, a novel technique is studied to compensate fibre nonlinearities in UWB systems: the multicarrier multiplexing, which tries to exploit the concept of symbol rate optimization. Finally, we assess the capabilities of this novel UWB SOA for data-centre interconnection applications with two experiments transmitting up to 113 Tbps data troughput in a 100-nm continuous bandwidth link over 100 km of fibre and then testings is stability with real-time line cards between two points of presence (POP)of Facebook deployed in the Paris area