<p dir="ltr">Porous silicon (po-Si) is a unique form of silicon (Si) that features tunable nanopores distributed throughout its bulk structure. While crystalline Si (c-Si) already boasts technological advantages, po-Si offers an additional key aspect with its large surface area relative to its small volume, making it highly conducive to surface chemistry. In this research, our focus centers on the design of a synaptic device based on po-Si, exploring its potential for neuromorphic hardware applications.</p><p><br></p><p dir="ltr">To begin, we delve into the analysis of several electrical properties of po-Si using density functional theory (ab initio/first principles) calculations. Notably, we discover the presence of intra-pore dangling states within the bandgap region of po-Si. Although po-Si is known for its higher bandgap compared to c-Si, resulting in low carrier density and increased resistance, the existence of these dangling states significantly impacts its electronic transport.</p><p><br></p><p dir="ltr">Additionally, we investigate the electric-field driven modulation of dangling bonds through controlled intra-pore Si-H bond dissociation. This modulation enables precise control over the density of dangling states, facilitating the tunability of po-Si conductance. Theoretically evaluating the current-voltage characteristics of our proposed po-Si based synaptic devices, we determine the potential range of obtainable conductivity.</p><p><br></p><p dir="ltr">Finally, we evaluate the performance by integrating porous silicon nanoelectronics devices into neural networks. These devices exhibit superior synaptic plasticity, faster response times, and reduced power consumption compared to other synapses. The research indicates that poroussilicon devices are highly effective in neuromorphic systems, paving the way for more efficient and scalable neural networks. These advancements have significant practical and cost-effective implications for a wide range of applications, including pattern recognition, machine learning, and artificial intelligence.</p><p><br></p><p dir="ltr">Overall, our analyses reveal that the integration of po-Si based synaptic devices into the neural fabric offers a path towards achieving significantly denser and more energy-efficient neuromorphic hardware. With its tunable properties, large surface area, and potential for controlled conductance, po-Si emerges as a promising candidate for the development of advanced silicon based nano-electronic devices tailored for neuromorphic computing. As we delve deeper into the potentials of po-Si, the era of cognitive computing, inspired by the elegance of bio-mimetic neural networks, edges closer to becoming a reality.<br><br></p>

Le silicium poreux est un biomatériau prometteur pour l'ingénierie tissulaire car il est non toxique et biorésorbable. Des modifications de surface permettent de contrôler sa vitesse de dégradation et peuvent favoriser l'adhésion cellulaire. Les cellules souches de la pulpe dentaire (DPSC) sont des cellules souches mésenchymateuses retrouvées dans la pulpe dentaire, à l'intérieur des dents, et constituent une source accessible de cellules souches. Regrouper les capacités de prolifération et différenciation des DPSC avec les propriétés morphologiques et biochimiques du pSi représente une approche intéressante pour des applications thérapeutiques de médecine régénératrice. Dans cette thèse, nous avons étudié le comportement de DPSC humaines sur des supports de pSi, avec des pores variant de quelques nanomètres à plusieurs centaines de nanomètres. Nous avons travaillé sur différentes fonctionnalisations chimiques afin d'optimiser l'adhésion cellulaire et de stabiliser le matériau : oxydation thermique, silanisation et hydrosilylation. L'adhésion, la prolifération et la différenciation osseuse ont été évaluées par microscopie à fluorescence, microscopie électronique à balayage, activité enzymatique, tests de prolifération (activité mitotique), immunofluorescence et spectroscopie FTIR. Le pSi avec des pores de 30 à 40 nm de diamètre s'est révélé être le plus approprié pour l'adhésion, la prolifération cellulaire et la différenciation ostéoblastique. De plus, la structure nanométrique et le relargage d'acide silicique par le pSi a démontré un effet positif sur l'induction osseuse et la formation d'une matrice minéralisée. Le pSi est donc apparu comme un matériau prometteur pour l'adhésion de cellules souches mésenchymateuses, que ce soit pour une transplantation immédiate in vivo ou pour expansion et différenciation in vitro. / Porous silicon (pSi) is a promising biomaterial for tissue engineering as it is both non-toxic and bioresorbable. Moreover, surface modification can offer control over the degradation rate of pSi and can also promote cell adhesion. Dental pulp stem cells (DPSC) are mesenchymal stem cells found within the teeth and constitute a readily source of stem cells. Coupling the good proliferation and differentiation capacities of DPSC with the textural and chemical properties of the pSi substrates provides an interesting approach for therapeutic use. In this thesis, the behavior of human DPSC is analyzed on pSi substrates presenting pore of various sizes, from few to hundreds nanometers. We investigated different chemical surface treatments, in order to enhance cell adhesion and stabilize the material: thermal oxidation, silanization and hydrosilylation. DPSC adhesion, proliferation and further osteodifferentiation were followed for up to 3 weeks by fluorescence microscopy, scanning electron microscopy (SEM), enzymatic activity assay, BrdU assay for mitotic activity, immunostaining and FTIR spectroscopy. Porous Silicon with pore size ranging from 30 to 40 nm was found to offer the best adhesion, the fastest growth rate for DPSC and the highest osteoinductive effect. Moreover, the pSi nanostructure and the release of silicic acid had a positive effect on precursor cells osteodifferentiation and mineralized matrix formation. Porous silicon appeared to be an appropriate biomaterial for mesenchymal stem cells adhesion and immediate in vivo transplantation, or for long term in vitro culture, for stem cells proliferation and differentiation.

Neste trabalho nós estudamos as propriedades de nanopartículas de Si hidrogenadas, limpas e com oxidação da superfície, como simulação do material Silício poroso. Para tal, desenvolvemos um procedimento para o cálculo da geometria, propriedades vibracionais e espectro ótico de sistemas semicondutores complexos, usando as técnicas semiempíricas de Química Quântica. As técnicas escolhidas foram completamente reparametrizadas para os átomos de Si e O, e assim apresentamos as novas parametrizações que chamamos AM1/Cristal e Zindo/Cristal. Contrariamente ao silício cristalino, o material poroso emite eficientemente luz no visível, com duas bandas, no vermelho-laranja e no verde. Esse comportamento tem sido atribuído ao confinamento quântico em estruturas nanocristalinas criadas pela porosidade, confinamente esse que deve ser responsável tanto pela eficiência da emissão, quanto pelo deslocamento do limiar ótico para energias mais altas. Nossos resultados para nanopartículas de diferentes diâmetros confirmam a cristalinidade das estruturas, e mostram um deslocamento para o azul do primeiro pico de absorção para partículas de diâmetro ~15 Å está em torno de 3 eV, muito mais altas do que a emissão vermelho-laranja. O estudo da relaxação estrutural no primeiro estado excitado mostra uma distorção forte e localizada, criando um defeito de superfície em que um átomo de hidrogênio coloca-se numa configuração de ponte Si-H-Si. Nessa configuração as partículas emitem numa região de energia mais baixa (vermelho-laranja), independentemente do diâmetro. A oxidação da superfície influencia muito pouco as propriedades óticas, e em particular não afeta a energia da linha de emissão. À luz destes resultados, associamos a atividade ótica do silício poroso a regiões nanocristalinas quase esféricas. A absorção (que varia consideravelmente em energia) e emissão no verde ocorreu no core cristalino, e a emissão vermelho-laranja na região de superfície, através desses defeitos fotocriados, sendo portanto fixa em energia. O deslocamento para o azul da absorção com a oxidação interpretamos como sendo devido à diminuição do diâmetro efetivo dos cristalinos presentes no material, e o decréscimo da intensidade de luminescência como devida a um enrijecimento da superfície oxidada, que reduz o número de sítios favoráveis à fotocriação de defeitos. / We study the properties of hydrogenated Si nanoparticles, also under surface oxidation, as a model-material to understand porous Silicon. To do that we developed a procedure designed to calculate geometries, vibrational properties and optical spectra for complex semiconductor systems, using semiempirical Quantum Chemistry techniques. The adopted techniques were thoroughly reparametrized for the Si and O atoms, and we thus present here the new parametrizations that we call AM1/Crystal and ZINDO/Crystal. Contrary to the bulk crystal, porous Si is known to emit visible light, efficiently, with bands in the red-orange and green regions. This behavior has been ascribed to quantum confinement in crystalline nanostructures created by the porosity, which should account both for the blue shift of the optical thereshold and for the emission efficiency. Our results for different nanoparticles confirm the crystallinity of the structures, and show a blue shift of the first absorption peak with decreasing diameter. However the absorption peak energy for nanoparticles with effective diameter around 15 Å lies around 3eV, much higher than the red-orange emission. A study of structural relaxation in the first excited state reveals a strong local distortion that creates a surface defect, in which an hydrogen atom \"bridges a pair of surface silicon atoms. In this Si-H-Si configuration the nanoparticles emit light of much lower energy (red-orange), which is virtually independent of diameter. Surface oxidation also has very little influence on the energy of the emitted light.Based on our results, we associate the optical activity of porous silicon to quasi- spherical nanocrystalline regions in the material. Both the absorption and green emission occur in the core of the crystallites, and shows blue-shift, with decreasing size; the red-orange luminescence occurs at the surface, through photo- generated defects, being thus pinned in energy. The blue shift of absorption with oxidation we interpret as being due to decrease in crystallite size, and the decrease in luminescence intensity as being due to \"hardening\" of the oxidized surface, which decreases the total number of sites for photogeneration of defects.

O presente projeto tem como objetivo contribuir para o desenvolvimento de dispositivos para monitoração de pressão dinâmica do ar. Para isso, foram produzidos filmes sensíveis a pressão baseados na detecção de concentração de oxigênio por meio de processos de emissão fotoluminescente das moléculas de Azul de Metileno (MB) e Platina Octaetilporfirina (PtOEP). Nesse sentido, foi estudado o comportamento da emissão fotoluminescente dessas moléculas em interação com o gás de oxigênio. A concentração de oxigênio (do ar) sobre superfície sensível depende da pressão dinâmica de superfície. Desta forma, monitorando a fotoluminescência dos dispositivos submetidos a diferentes concentrações de oxigênio, pode-se determinar uma relação entre a pressão pontual da superfície analisada e a intensidade de emissão fotoluminescente do filme. Os dispositivos de monitoração de pressão dinâmica são constituídos de um filme de estado sólido contendo as moléculas sensíveis. Foram utilizados como substratos hospedeiros para o Azul de Metileno e para a Octaetilporfirina de Platina, os filmes de silício poroso oxidado e Poliestireno, respectivamente. É proposto um arranjo experimental que utiliza um fluorímetro para caracterizar as amostras produzidas e uma câmara de fluxo de gases. Os dispositivos apresentaram elevada sensibilidade e evidenciaram o potencial para desenvolvimento e integração de sensores baseados no silício poroso à microeletrônica. / This project aims to contribute to the development of devices for monitoring dynamic pressure of the air. In this sense, films were produced based on pressure-sensitive detection of oxygen concentration through processes of photoluminescence emission from the molecules of methylene blue (MB) and platinum octaethylporphyrin (PtOEP). Accordingly, it was studied the behavior of the photoluminescence emission of these molecules in interaction with the oxygen gas. The concentration of oxygen (of the air composition) on the sensitive surface depends on the surface dynamic pressure. Thus, monitoring the photoluminescence of the devices under different oxygen concentrations, it can be determined a relationship between the punctual pressure of the tested surface and the photoluminescence emission intensity of the film. The devices for monitoring dynamic pressure are made of a solid state film containing the sensitive molecules. Oxidized porous silicon and polystyrene films were used as hosts for the Methylene Blue and for the Platinum Octaethylporphyrin, respectively. It was proposed an experimental setup that uses a spectrofluorophotometer and a gas flow chamber to characterize the produced samples. The devices showed high sensitivity and potential for development and integration of the sensors based on porous silicon for microelectronics.

Neste trabalho nós estudamos as propriedades de nanopartículas de Si hidrogenadas, limpas e com oxidação da superfície, como simulação do material Silício poroso. Para tal, desenvolvemos um procedimento para o cálculo da geometria, propriedades vibracionais e espectro ótico de sistemas semicondutores complexos, usando as técnicas semiempíricas de Química Quântica. As técnicas escolhidas foram completamente reparametrizadas para os átomos de Si e O, e assim apresentamos as novas parametrizações que chamamos AM1/Cristal e Zindo/Cristal. Contrariamente ao silício cristalino, o material poroso emite eficientemente luz no visível, com duas bandas, no vermelho-laranja e no verde. Esse comportamento tem sido atribuído ao confinamento quântico em estruturas nanocristalinas criadas pela porosidade, confinamente esse que deve ser responsável tanto pela eficiência da emissão, quanto pelo deslocamento do limiar ótico para energias mais altas. Nossos resultados para nanopartículas de diferentes diâmetros confirmam a cristalinidade das estruturas, e mostram um deslocamento para o azul do primeiro pico de absorção para partículas de diâmetro ~15 Å está em torno de 3 eV, muito mais altas do que a emissão vermelho-laranja. O estudo da relaxação estrutural no primeiro estado excitado mostra uma distorção forte e localizada, criando um defeito de superfície em que um átomo de hidrogênio coloca-se numa configuração de ponte Si-H-Si. Nessa configuração as partículas emitem numa região de energia mais baixa (vermelho-laranja), independentemente do diâmetro. A oxidação da superfície influencia muito pouco as propriedades óticas, e em particular não afeta a energia da linha de emissão. À luz destes resultados, associamos a atividade ótica do silício poroso a regiões nanocristalinas quase esféricas. A absorção (que varia consideravelmente em energia) e emissão no verde ocorreu no core cristalino, e a emissão vermelho-laranja na região de superfície, através desses defeitos fotocriados, sendo portanto fixa em energia. O deslocamento para o azul da absorção com a oxidação interpretamos como sendo devido à diminuição do diâmetro efetivo dos cristalinos presentes no material, e o decréscimo da intensidade de luminescência como devida a um enrijecimento da superfície oxidada, que reduz o número de sítios favoráveis à fotocriação de defeitos. / We study the properties of hydrogenated Si nanoparticles, also under surface oxidation, as a model-material to understand porous Silicon. To do that we developed a procedure designed to calculate geometries, vibrational properties and optical spectra for complex semiconductor systems, using semiempirical Quantum Chemistry techniques. The adopted techniques were thoroughly reparametrized for the Si and O atoms, and we thus present here the new parametrizations that we call AM1/Crystal and ZINDO/Crystal. Contrary to the bulk crystal, porous Si is known to emit visible light, efficiently, with bands in the red-orange and green regions. This behavior has been ascribed to quantum confinement in crystalline nanostructures created by the porosity, which should account both for the blue shift of the optical thereshold and for the emission efficiency. Our results for different nanoparticles confirm the crystallinity of the structures, and show a blue shift of the first absorption peak with decreasing diameter. However the absorption peak energy for nanoparticles with effective diameter around 15 Å lies around 3eV, much higher than the red-orange emission. A study of structural relaxation in the first excited state reveals a strong local distortion that creates a surface defect, in which an hydrogen atom \"bridges a pair of surface silicon atoms. In this Si-H-Si configuration the nanoparticles emit light of much lower energy (red-orange), which is virtually independent of diameter. Surface oxidation also has very little influence on the energy of the emitted light.Based on our results, we associate the optical activity of porous silicon to quasi- spherical nanocrystalline regions in the material. Both the absorption and green emission occur in the core of the crystallites, and shows blue-shift, with decreasing size; the red-orange luminescence occurs at the surface, through photo- generated defects, being thus pinned in energy. The blue shift of absorption with oxidation we interpret as being due to decrease in crystallite size, and the decrease in luminescence intensity as being due to \"hardening\" of the oxidized surface, which decreases the total number of sites for photogeneration of defects.

Le Silicium est le deuxième élément le plus abondant dans la croûte terrestre après l’oxygène. Il est produit par voie métallurgique dans un four à arc électrique, le quartz est réduit en présence de réducteurs (charbon de bois, houille et coke de pétrole). Le silicium métallurgique est principalement utilisé dans la métallurgie comme élément d’alliage, dans la chimie et l’industrie solaire. Le prix du Silicium est fonction de sa pureté. Les travaux de cette thèse se divisent en deux parties l’utilisation du Silicium Métallurgique (99% Si) pour le stockage de l’hydrogène, et la photoluminescence du ferrosilicium (disiliciure de fer) de qualité métallurgique. Des substrats de silicium métallurgique ont été soumis à une anodisation électrochimique dans une solution à base d’acide fluorhydrique. Le silicium poreux nanostructuré obtenu est légèrement différent du silicium poreux issu de substrat de silicium de qualité électronique de même résistivité. L’influence des principaux paramètres sur la génération de l’hydrogène : la porosité, la concentration, le volume et la température ont fait l’objet d’une étude détaillée. Le silicium poreux produit à partir de silicium métallurgique est un matériau de stockage d’hydrogène. Des substrats de disiliciure de fer de qualité métallurgique ont été soumis à une anodisation électrochimique. Le composé obtenu est du disiliciure de fer nanostructuré avec du silicium résiduel, ce produit est recouvert de fluorosilicate de fer hexahydraté qui a la particularité d’être luminescent. Il s’agit à ce jour de la première anodisation du disiliciure de fer, un mécanisme de gravure a été proposé et l’influence des principaux paramètres d’anodisation sur les propriétés de photoluminescence a été évaluée. / Silicon is the second most abundant element in the Earth crust after oxygen. Its use in metallurgy, building and electronic industry requires a huge fabrication level. Depending on the contamination level allowed, the price of this material varies in the orders of magnitude. This thesis focuses on the use of dirtiest metallurgical grade silicon and iron disilicide substrates for hydrogen storage and photoluminescence applications. The initial substrates were subjected to electrochemical etching in hydrofluoric acid-containing solutions. Anodization of metallurgical grade silicon substrate produces nanostructured porous silicon with somewhat shifted parameters (comparing with electronic grade porous silicon with the same resistivity), as it was studied in this thesis in details. It was shown, that metallurgical grade porous silicon can be applied as hydrogen storage material. Hydrogen generation is studied here based on the influences of some technically critical parameters: porosity, alkali concentration, volume and temperature. Electrochemical treatment of metallurgical grade iron disilicide substrates produces luminescent iron fluorosilicate hexahydrate, covering the residual nanostructured iron disilicide/silicon. Here, the influence of anodization parameters on photoluminescent properties is studied. Also, etching mechanism is proposed as for the new material never anodized.

Diatoms possess characteristics such as abundance, diversity, and high reproductivity, which make their nano-structured frustules (diatom frustules) attractive for a wide range of applications. To overcome the limitation of their silica based frustule composition, diatom frustules have been converted into a variety of materials including silicon, silicon carbide, silver, gold, palladium and carbon in the present study. The compositions and the extent of shape preservation of the replicas are examined and evaluated with different characterization methods such as X-ray diffraction, SEM, TEM and FTIR analyses. These replicas still retained the complex 3D structures and nano-scaled features of the starting diatom frustules. Some properties and possible applications of converted materials are explored and the kinetics and thermodynamics related to the successful replications (conversions) are also studied and discussed.

Cette thèse a été consacrée à l’étude du Si poreux comme matériaux à faible conductivité thermique (k) pour application aux dispositifs thermoélectriques à base de Si. D’autres paramètres thermoélectriques, comme par exemple le coefficient Seebeck de ce matériau, ont été également étudiés.Si poreux est un matériau complexe composé de nanostructures de Si séparées de vide. Quand la porosité est élevée, sa conductivité thermique est bien inférieure à celle de Si cristallin. Nous avons étudié la conductivité thermique de Si poreux de différentes morphologies et porosités dans la gamme de températures 4.2-350K. Les mesures à T<20K sont les premières dans la bibliographie et ont montré une saturation de k en fonction de T pour ces températures. A des températures supérieures à 20K, k augmente régulièrement avec la température. La dépendance de température de k de Si poreux a été interprétée en considérant des modèles théoriques, basées sur la nature “fractal” de Si poreux. Nous avons calculé la dimension fractale de Si poreux par des images de microscopie électronique à balayage (SEM) et l’algorithme de “box counting”.Deux méthodes différentes ont été utilisées pour mesurer k: la méthode à courant direct (dc) combinée avec une analyse FEM et la méthode 3ω. Nous avons proposé deux approches améliorées pour extraire k du signal de potentiel 3ω en fonction de la fréquence. La première considère l’accord des résultats expérimentaux avec la solution asymptotique intégrale de la formule de Cahill, et la seconde fait une analyse des résultats expérimentaux en solvant l’équation temporelle de transfert de chaleur par des éléments finis. Plus précise est la méthode 3ω combinée avec des éléments finis. Les résultats correspondants sont en bon accord avec ceux obtenus par la méthode dc.Nous avons aussi étudié le Si poreux comme matériau thermoélectrique. Dans ce cas, le Si poreux peut être intéressant si il a une faible porosité, car le matériau à haute porosité est très résistive. Dans ce but, nous avons déterminé le coefficient Seebeck (S) des membranes de Si poreux de différentes porosités dans la gamme 40-84%, en utilisant un dispositif de mesure spécialement développé à cet effet. Pour des échantillons de porosité 51%, la valeur de coefficient S est de 1mV/K, bien supérieure à celle le Si cristallin. La dépendance de S de la porosité n’est pas monotone, et ceci est attribué à une combinaison des effets de filtrage d’énergie, des collisions des phonons et interactions phonon-porteurs électriques. Les résultats obtenus sont basées sur des mesures de photoluminescence (PL) et observations microscopiques à transmission (TEM). Nous avons enfin conclue que, malgré le coefficient S très élevé, le Si poreux n’est pas adéquat comme matériau thermoélectrique à cause de sa faible conductivité électrique, qui diminue en augmentant la porosité à cause de la résultante déplétion de porteurs.Nous avons aussi étudié des films minces polycristallins dopés avec du Bore. Ces films sont très intéressants comme matériaux thermoélectrique, car ils sont compatibles avec les procédés de fabrication des circuits intégrés de Si. Leur performance thermoélectrique est améliorée par diminution de la taille des grains. Des films minces polycristallins d’épaisseur entre 100 et 500nm ont été étudiés. Tous leurs paramètres thermoélectriques ont été mesurés et nous avons trouvé que le facteur de performance thermoélectrique zT augmente d’un facteur 3 en diminuant l’épaisseur de 500 à 100nm ceci étant attribué à la diminution de la taille des grains dans les films, conduisant à zT = 0.033, qui est la meilleure valeur reporté dans la littérature.Ce résultat compétitif augmente le potentiel d’utilisation des films polycristallins dans des dispositifs thermoélectriques efficaces, compatibles à la technologie de Si. / This thesis is devoted to the thermal conductivity and other thermoelectric properties of porous silicon (PSi) and thin polycrystalline Si films (thickness: 100-500 nm).PSi is a complex material composed of a Si skeleton of interconnected nanowires and dots, separated by voids. When it is highly porous, its thermal conductivity is very low, even below that of the amorphous Si. This makes it a good material for use as a thermal isolation platform on the Si wafer. In addition, its Seebeck coefficient is much higher than that of bulk c-Si.We studied k of PSi layers with different morphologies and porosities, in the temperature range 4.2-350K. The measurements below 20K are the first reported in the literature. A plateau-like dependence on temperature was observed for T below 20K, while above this temperature a monotonic increase with T is observed. The observed behaviour was interpreted using known theoretical models, based mainly on the fractal nature of PSi. PSi was characterized as a fractal material by calculating its fractal dimension using SEM images and the box counting algorithm.Two different methods were used to determine porous Si thermal conductivity: the DC method combined with FEM analysis and the 3ω method. Concerning the 3ω method, two improved approaches were proposed for extracting k from the 3ω voltage as a function of frequency: the first uses a fitting of the experimental data to the asymptotic solution of the Cahill’s integral formula, and the second is based on the analysis of the experimental data by combining them with a solution of the transient heat transfer equation using FEM analysis. The results in this second case were more accurate and in very good agreement with the DC method.We also measured the Seebeck coefficient (S) of PSi membranes with porosities 40-84% using a home-built setup, which was fabricated, calibrated and tested within this thesis. A value as high as 1mV/K was obtained for the 51% porosity sample. An anomalous porosity dependence of S was obtained, which was attributed to the interplay between energy filtering, phonon scattering and phonon drag effects. The results were explained by combining them with PL and TEM measurements, used for the determination of nanocrystal sizes. We concluded that, despite of the extremely low k and the high S of PSi, the material with the studied high porosities is not adequate for use as a “good thermoelectric” material, because of its significantly low electrical conductivity, which decreases with increasing porosity, resulting from carrier depletion during formation.We also studied the thermoelectric properties of thin, boron-doped, polycrystalline silicon films, which are much more attractive for use as Si-based thermoelectrics than porous Si. Their thermoelectric performance is improved by decreasing film thickness, due to a decrease in polysilicon grain size. Thin films with thickness between 100-500nm were investigated. We measured their thermal conductivity, resistivity and Seebeck coefficient and extracted their thermoelectric figure of merit, which showed threefold increase by reducing film thickness down to 100nm. A value as high as 0.033 was achieved, which is the highest reported in the literature so far for boron-doped polysilicon films at room temperature. This increase is attributed to a decrease in the grain size of the material. The obtained value shows the interest of nanocrystalline Si films for integration in efficient Si-based thermoelectric generators, compatible with CMOS processing.

L’intégration monolithique de filtres et de diodes contre les décharges électrostatiques sur silicium est une solution à bas cout et fiable pour protéger les interfaces de transfert de données des appareils nomades. La formation de caissons isolants de silicium poreux sous les filtres permet d’améliorer en partie leur performance. Cette thèse avait pour but de poursuivre l’intégration de démonstrateurs RF sur silicium poreux et de proposer des voies de fonctionnalisation du matériau en vue d’optimiser les caractéristiques des filtres. Tout d’abord, les configurations bénéfiques d’intégration de caissons poreux à un filtre de mode commun (ECMF) ont été étudiées. Ensuite, une optimisation de l’étape d’oxydation post-anodisation a été réalisée afin d’améliorer la qualité et la stabilité de l’isolation électrique. Pour cela, les mécanismes d’oxydation, les propriétés chimiques et les propriétés électriques du silicium mésoporeux oxydé ont été mises en perspective avec la nature du traitement appliqué. Enfin, l’insertion dans les pores de nanoparticules ferromagnétiques de forte perméabilité a été menée dans le but d’augmenter la densité d’inductance par unité de surface. / Monolithic integration of interference filters and protection diodes on silicon is a viable and mature technology used to protect high-speed serial interfaces of nomadic devices. To enhance the filters performance, porous silicon can be formed by anodization specifically underneath the filter area. This thesis aimed to pursue the integration of RF prototypes on porous silicon and also to suggest strategies of material functionalization in order to optimize the filter characteristics. First, various configurations of common-mode filters were integrated on porous silicon and their performances were compared. Then, the post-anodization oxidation step has been optimized in order to provide the most efficient and stable electrical isolation. The oxidation mechanisms were discussed. The surface chemistry of porous silicon and its electrical behavior have been put in perspective with the oxidation treatments. Finally, this work suggested experimental methods to synthesize and characterize a ferromagnetic porous silicon-based nanocomposite for the improvement of the inductance density per unit area.

Ces travaux de thèse portent sur la fabrication de via traversants conducteurs, brique technologique indispensable pour l’intégration des composants microélectroniques en 3 dimensions. Pour ce faire, une voie « tout-électrochimique » a été explorée en raison de son faible coût de fabrication par rapport aux techniques par voie chimique sèche. Ainsi, la gravure de macropores ordonnés traversants a été réalisée par anodisation du silicium en présence d’acide fluorhydrique puis leur remplissage de cuivre par dépôt électrochimique. L’objectif est de faire du silicium macroporeux une alternative crédible à la gravure sèche (DRIE) pour la structuration du silicium.Les conditions de gravure de matrices de macropores ordonnés traversants ont été étudiées à la fois dans des substrats silicium de type n et p faiblement dopés. La composition de l’électrolyte ainsi que le motif des matrices ont été optimisés afin de garantir la gravure de via traversants de forte densité et à facteur de forme élevé. Une fois gravés, les via traversant ont été remplis de cuivre. En optimisant ces paramètres une résistance minimale égale à 32 mΩ/via (soit 1,06 fois la résistivité théorique du cuivre à 20°C) a été mesurée. / These thesis works deal with the achievement of Through Silicon Via (TSV) essential technological issue for microelectronic device 3D integration. For this purpose, we opted for a “full-electrochemical” way of TSV production because of lower fabrication costs as compared to dry etching and deposition techniques. Indeed, ordered through silicon macropores were carried out by silicon anodization in hydrofluoric acid-containing solution and then filled by copper electrochemical deposition. The main objective is to determine if the macroporous silicon arrays can be a viable alternative as Deep Reactive Ion Etching (DRIE).The etching parameters of through silicon macropore arrays were studied both in low-doped n- and p-type silicon. The electrolyte composition as well as the density of the initiation sites was optimized to enable the growth of high aspect ratio, high density through silicon ordered macropores. After silicon anodization, through via were filled with copper. By optimizing the copper deposition parameters (bath composition and applied potential), the resistance per via was measured equal to 32 mΩ (i.e. 1.06 times higher than the theoretical copper bulk resistivity).