Ce mémoire est consacré au développement et à l'application d'une méthode permettant de caractériser finement les propriétés physiques (densité d'hélium, pression, diamètre, morphologie) de bulles d'hélium de taille nanométrique pour in fine améliorer la compréhension du comportement de ces nano-systèmes. L'approche que nous avons choisie est basée sur la spectroscopie de pertes d'énergie des électrons et l'acquisition de spectres images en microscopie électronique en transmission filtrée en énergie. Les différentes étapes d'acquisition, de correction des aberrations, et de traitement des spectres sont détaillées. L'erreur sur la mesure est estimée, et des améliorations potentielles de la méthode sont discutées. Nous montrons de plus que cette approche permet non seulement de dépasser les limites imposées par la microscopie électronique en transmission à balayage habituellement utilisée, mais aussi d'aller au-delà, en terme de statistique notamment. Nous appliquons ensuite notre méthode pour déterminer les propriétés physiques de bulles d'hélium dans le silicium, lors de recuits thermiques in situ dans le microscope. L'évolution des caractéristiques morphologiques des bulles est mise en rapport avec la variation de la densité d'hélium qu'elles contiennent suite à ces recuits. Les valeurs de densité et de pression obtenues sont comparées aux valeurs disponibles dans la littérature par des méthodes expérimentales ou numériques. Enfin, le transfert de notre méthode pour l'étude de bulles dans d'autres matrices (germanium, carbure de silicium, euxénite) est discuté. / This thesis is dedicated to the development and application of a method allowing for the fine characterization of the physical properties (density, diameter, pressure and morphology) of helium bubbles at the nanometric scale, to eventually improve the understanding of the behavior of these nano-systems. The chosen approach is based on electron energy loss spectroscopy and the acquisition of spectral images by energy-filtered transmission electron microscopy. The acquisition, aberration correction, and data analysis steps are detailed. The measurement error is estimated, and potential improvements are discussed. We additionally show that this method is not only able to overcome the limits imposed by commonly used scanning transmission electron microscopy, but also to go further, notably in a statistical way. We then apply our method to determine the physical properties of helium bubbles in silicon, during in situ thermal annealing in the microscope. The evolution of the morphological characteristics of the bubbles is put in relation with the variation of the density of the helium contained following these annealings. The density and pressure values are compared to those available in the litterature through experimental and numerical methods. Finally, the translation of this method towards the study of bubbles in other matrices (germanium, silicon carbide, euxenite) is discussed.

Ces travaux de doctorat concernent le développement de la tomographie électronique appliquée à la nano-caractérisation tridimensionnelle de dispositifs à semi-conducteurs et de matériaux pour la micro et la nanoélectronique. Les contributions les plus significatives de ces travaux sont (i) l'exploration et l'application de différents modes de contraste en microscopie électronique à transmission (TEM) pour des applications spécifiques liées au semi-conducteurs et (ii) l'investigation de nouvelles pistes pour améliorer encore la résolution spatiale, en particulier en adaptant les schémas d'acquisition en tomographie. Le TEM en balayage (STEM), basé sur des mesures annulaires aux forts angles et en champ sombre (HAADF) a été mis en œuvre pour observer des dopants dont le numéro atomique est typiquement largement supérieur à celui de la matrice (en silicium), et nous avons combiné le TEM résolu en énergie (EFTEM) dans un régime de faible perte d'énergie des électrons avec les techniques de tomographie afin de reconstruire les spectres de perte d'énergie locaux, en chaque voxel. La tomographie double-axe a été expérimentalement mise en œuvre pour améliorer la résolution spatiale, et le potentiel de la tomographie à axe multiple a été démontré, grâce aux simulations. Enfin, des algorithmes de reconstruction basés sur la minimisation de la variation totale ont été appliqués à la tomographie électronique. Les analyses effectuées comprennent les transistors triple-grille, les nanofils III-V, les capacités à base de nanofils de silicium et le silicium sur-dopé au sélénium, un matériau utilisé pour des applications optoélectroniques. / In this thesis electron tomography is developed and applied as a tool for three-dimensional nanoscale characterization of semiconductor materials and devices. The major contributions of this thesis are the exploration and application of transmission electron microscopy (TEM) contrast techniques for specific semiconductor applications and the exploration of routes towards improving spatial resolution, in particular by adapting tomographic acquisition schemes. As contrast techniques we apply high-angle annular dark-field (HAADF) scanning TEM (STEM) for investigations of heavy dopants in a lighter environment and we combine spectral low-loss energy-filtered TEM (EFTEM) with tomography and explore the features of reconstructed low-loss spectra. For resolution improvement we experimentally apply dual-axis electron tomography and investigate the potential of multiple-axis tomography based on simulations. Furthermore reconstruction algorithms based on totalvariation minimization are applied to electron tomography. Samples investigated in this work include tri-gate transistors, III-V nanowire heterostructures and silicon nanowire based capacitors as well as selenium-hyperdoped silicon, a material for optoelectronic applications.

The demand for electrical energy is steadily increasing. Highly efficient organic solar cells based on mixed, strongly absorbing organic molecules convert sunlight into electricity and, thus, have the potential to contribute to the worlds energy production. The continuous development of new materials during the last decades lead to a swift increase of power conversion efficiencies (PCE) of organic solar cells, recently reaching 12%. Despite these breakthroughs, the usage of highly complex organic molecules blended together to form a self-organised absorber layer results in complicated morphologies that are poorly understood. However, the morphology has a tremendous impact on the photon-to-electron conversion, affecting all processes ranging from light absorption to charge carrier extraction. This dissertation studies the role of phase-separation of the self-organised thin film blend layers utilized in organic solar cells. On the molecular scale, we manipulate the phase-separation, using different molecule combinations ranging from the well-known ZnPc:C 60 blend layers to highly efficient oligothiophene:C60 blend layers. On the macroscopic scale, we shape the morphology by depositing the aforementioned blend layers on differently heated substrates (in-vacuo substrate temperature, Tsub). To characterise the manufactured blend layers, we utilize high resolution microscopy techniques such as photoconductive atomic force microscopy, different electron microscopic techniques, X-ray microscopy etc., and various established and newly developed computational simulations to rationalise the experimental findings. This multi-technique, multi-scale approach fulfils the demands of several scientific articles to analyse a wide range of length scales to understand the underlying optoelectronic processes. Varying the mixing ratio of a ZnPc:C60 blend layer from 2:1 to 6:1 at fixed in vacuo substrate temperature results in a continuous increase of surface roughness, decrease of short-circuit current, and decrease of crystallinity. Additionally performed density functional theory calculations and 3D drift-diffusion simulations explain the observed crystalline ZnPc nanorod formation by the presence of C60 in the bulk volume and the in turn lowered recombination at crystalline ZnPc nanorods. Moving to oligothiophene:C60 blend layers used in highly efficient organic solar cells deposited at elevated substrate temperatures, we find an increase of phase-separation, surface roughness, decrease of oligothiophene-C60 contacts, and reduced disorder upon increasing Tsub from RT (PCE=4.5%) to 80 °C (PCE=6.8%). At Tsub =140 °C, we observe the formation of micrometer-sized aggregates on the surface resulting in inhomogeneous light absorption and charge carrier extraction, which in turn massively lowers the power conversion efficiency to 1.9%. Subtly changing the molecular structure of the oligothiophene molecule by attaching two additional methyl side chains affects the thin film growth, which is also dependent on the substrate type. In conclusion, the utilized highly sensitive characterisation methods are suitable to study the impact of the morphology on the device performance of all kinds of organic electronic devices, as we demonstrate for organic blend layers. At the prototypical ZnPc:C60 blend, we discovered a way to grow ZnPc nanorods from the blend layer. These nanorods are highly crystalline and facilitate a lowered charge carrier recombination which is highly desirable in organic solar cells. The obtained results at oligothiophene: C60 blends clearly demonstrate the universality of the multi-technique approach for an in-depth understanding of the fragile interplay between phase-separation and phase-connectivity in efficient organic solar cells. Overall, we can conclude that both molecular structure and external processing parameters affect the morphology in manifold ways and, thus, need to be considered already at the synthesis of new materials.

The demand for electrical energy is steadily increasing. Highly efficient organic solar cells based on mixed, strongly absorbing organic molecules convert sunlight into electricity and, thus, have the potential to contribute to the worlds energy production. The continuous development of new materials during the last decades lead to a swift increase of power conversion efficiencies (PCE) of organic solar cells, recently reaching 12%. Despite these breakthroughs, the usage of highly complex organic molecules blended together to form a self-organised absorber layer results in complicated morphologies that are poorly understood. However, the morphology has a tremendous impact on the photon-to-electron conversion, affecting all processes ranging from light absorption to charge carrier extraction. This dissertation studies the role of phase-separation of the self-organised thin film blend layers utilized in organic solar cells. On the molecular scale, we manipulate the phase-separation, using different molecule combinations ranging from the well-known ZnPc:C 60 blend layers to highly efficient oligothiophene:C60 blend layers. On the macroscopic scale, we shape the morphology by depositing the aforementioned blend layers on differently heated substrates (in-vacuo substrate temperature, Tsub). To characterise the manufactured blend layers, we utilize high resolution microscopy techniques such as photoconductive atomic force microscopy, different electron microscopic techniques, X-ray microscopy etc., and various established and newly developed computational simulations to rationalise the experimental findings. This multi-technique, multi-scale approach fulfils the demands of several scientific articles to analyse a wide range of length scales to understand the underlying optoelectronic processes. Varying the mixing ratio of a ZnPc:C60 blend layer from 2:1 to 6:1 at fixed in vacuo substrate temperature results in a continuous increase of surface roughness, decrease of short-circuit current, and decrease of crystallinity. Additionally performed density functional theory calculations and 3D drift-diffusion simulations explain the observed crystalline ZnPc nanorod formation by the presence of C60 in the bulk volume and the in turn lowered recombination at crystalline ZnPc nanorods. Moving to oligothiophene:C60 blend layers used in highly efficient organic solar cells deposited at elevated substrate temperatures, we find an increase of phase-separation, surface roughness, decrease of oligothiophene-C60 contacts, and reduced disorder upon increasing Tsub from RT (PCE=4.5%) to 80 °C (PCE=6.8%). At Tsub =140 °C, we observe the formation of micrometer-sized aggregates on the surface resulting in inhomogeneous light absorption and charge carrier extraction, which in turn massively lowers the power conversion efficiency to 1.9%. Subtly changing the molecular structure of the oligothiophene molecule by attaching two additional methyl side chains affects the thin film growth, which is also dependent on the substrate type. In conclusion, the utilized highly sensitive characterisation methods are suitable to study the impact of the morphology on the device performance of all kinds of organic electronic devices, as we demonstrate for organic blend layers. At the prototypical ZnPc:C60 blend, we discovered a way to grow ZnPc nanorods from the blend layer. These nanorods are highly crystalline and facilitate a lowered charge carrier recombination which is highly desirable in organic solar cells. The obtained results at oligothiophene: C60 blends clearly demonstrate the universality of the multi-technique approach for an in-depth understanding of the fragile interplay between phase-separation and phase-connectivity in efficient organic solar cells. Overall, we can conclude that both molecular structure and external processing parameters affect the morphology in manifold ways and, thus, need to be considered already at the synthesis of new materials.

In this thesis 9-12 % Cr steels, used for high-temperaturecomponents in fossil-fired power plants are considered. Thecreep strength of thees steels depend on their microstructurethat consists of a matrix of tempered lath martensite withdensely distributed precipitates. The mechanical properties arestrongly influenced by precipitates present in the matrix andthe more densely distributed they are the higher is thehardening effect. These particles nucleate, grow and coarsenduring use in power plants, leading to a degradation ofmechanical properties. In this thesis the nucleation andcoarsening behaviour of the precipitates in the Cr steels aresimulated by new models and the results are compared withtransmission electron microscopy (TEM) observations of testedmaterials. A model of the nucleation of MX, which mainly isvanadium-nitrides, is presented. MX precipitates are assumed tonucleate on dislocations during annealing. The model takes intoaccount the full multicomponent thermodynamical behaviour ofthe system as well as the strain energy caused by a puredilatoric strain and the shape of the nucleus. The calculationsyield a critical size and shape represented by an oblatespheroid with a radius of a few nanometers and a thickness ofan atomic layer. This is in agreement with TEM observations ofsupercritical particles. A new coarsening model, which takes into account themulticomponent effects, is presented and it has been tested ondifferent systems to validate the model.     Four different carbides, Cr7C3, Mo6C, VC and NbC, in austenitic matrix as well asγ ' in ternary Ni-base super-alloy systems (Ni-Al-Mo)were compared with measurements from literature.     TiC-particles in austenitic stainless steel, ASTM 316Ti,were considered. The simulations were performed taking intoaccount iron and 7 alloying elements. The measurements wereperformed with TEM on samples that had been heat-treated at900° C.     MX and M23C6in a 9 % Cr steel were investigated with energyfiltering transmission electron microscopy (EFTEM) andcompared with simulations. The used samples had been heattreated for various periods of time at 600 and 650° Cfor up to 26 000 h. The agreement of the simulations with the experiments wasgood in all cases when reasonable values of the interfacialenergy, the only adjustable parameter, were chosen. Coarsening simulations were also performed to investigatethe influence of changes in composition on the coarsening rate.For MX, in a 9 % Cr steel, the coarsening rate is almostindependent of the V/Nb ratio but highly dependent on theN-content. Also the effect on the coarsening rate for M23C6by adding Co to a Cr steel was investigated bysimulation. Co is known to increase the resistance totempering. The results show that a final average radius of thecarbides after 30 000 h at 600° C decreases with 30 % witha Co addition of 10 mass %. <b>Keywords:</b>Cr steels, nucleation, coarsening, model,DICTRA, precipitates, carbides, carbo-nitrides, MX, VN, M23C6, TiC, TEM, EFTEM, Curie-temperature

<p>In this thesis 9-12 % Cr steels, used for high-temperaturecomponents in fossil-fired power plants are considered. Thecreep strength of thees steels depend on their microstructurethat consists of a matrix of tempered lath martensite withdensely distributed precipitates. The mechanical properties arestrongly influenced by precipitates present in the matrix andthe more densely distributed they are the higher is thehardening effect. These particles nucleate, grow and coarsenduring use in power plants, leading to a degradation ofmechanical properties. In this thesis the nucleation andcoarsening behaviour of the precipitates in the Cr steels aresimulated by new models and the results are compared withtransmission electron microscopy (TEM) observations of testedmaterials.</p><p>A model of the nucleation of MX, which mainly isvanadium-nitrides, is presented. MX precipitates are assumed tonucleate on dislocations during annealing. The model takes intoaccount the full multicomponent thermodynamical behaviour ofthe system as well as the strain energy caused by a puredilatoric strain and the shape of the nucleus. The calculationsyield a critical size and shape represented by an oblatespheroid with a radius of a few nanometers and a thickness ofan atomic layer. This is in agreement with TEM observations ofsupercritical particles.</p><p>A new coarsening model, which takes into account themulticomponent effects, is presented and it has been tested ondifferent systems to validate the model.</p><p>    Four different carbides, Cr₇C₃, Mo₆C, VC and NbC, in austenitic matrix as well asγ ' in ternary Ni-base super-alloy systems (Ni-Al-Mo)were compared with measurements from literature.</p><p>    TiC-particles in austenitic stainless steel, ASTM 316Ti,were considered. The simulations were performed taking intoaccount iron and 7 alloying elements. The measurements wereperformed with TEM on samples that had been heat-treated at900° C.</p><p>    MX and M₂₃C₆in a 9 % Cr steel were investigated with energyfiltering transmission electron microscopy (EFTEM) andcompared with simulations. The used samples had been heattreated for various periods of time at 600 and 650° Cfor up to 26 000 h.</p><p>The agreement of the simulations with the experiments wasgood in all cases when reasonable values of the interfacialenergy, the only adjustable parameter, were chosen.</p><p>Coarsening simulations were also performed to investigatethe influence of changes in composition on the coarsening rate.For MX, in a 9 % Cr steel, the coarsening rate is almostindependent of the V/Nb ratio but highly dependent on theN-content. Also the effect on the coarsening rate for M₂₃C₆by adding Co to a Cr steel was investigated bysimulation. Co is known to increase the resistance totempering. The results show that a final average radius of thecarbides after 30 000 h at 600° C decreases with 30 % witha Co addition of 10 mass %.</p><p><b>Keywords:</b>Cr steels, nucleation, coarsening, model,DICTRA, precipitates, carbides, carbo-nitrides, MX, VN, M₂₃C₆, TiC, TEM, EFTEM, Curie-temperature</p>

Ces travaux de doctorat concernent le développement de la tomographie électronique appliquée à la nano-caractérisation tridimensionnelle de dispositifs à semi-conducteurs et de matériaux pour la micro et la nanoélectronique. Les contributions les plus significatives de ces travaux sont (i) l'exploration et l'application de différents modes de contraste en microscopie électronique à transmission (TEM) pour des applications spécifiques liées au semi-conducteurs et (ii) l'investigation de nouvelles pistes pour améliorer encore la résolution spatiale, en particulier en adaptant les schémas d'acquisition en tomographie. Le TEM en balayage (STEM), basé sur des mesures annulaires aux forts angles et en champ sombre (HAADF) a été mis en œuvre pour observer des dopants dont le numéro atomique est typiquement largement supérieur à celui de la matrice (en silicium), et nous avons combiné le TEM résolu en énergie (EFTEM) dans un régime de faible perte d'énergie des électrons avec les techniques de tomographie afin de reconstruire les spectres de perte d'énergie locaux, en chaque voxel. La tomographie double-axe a été expérimentalement mise en œuvre pour améliorer la résolution spatiale, et le potentiel de la tomographie à axe multiple a été démontré, grâce aux simulations. Enfin, des algorithmes de reconstruction basés sur la minimisation de la variation totale ont été appliqués à la tomographie électronique. Les analyses effectuées comprennent les transistors triple-grille, les nanofils III-V, les capacités à base de nanofils de silicium et le silicium sur-dopé au sélénium, un matériau utilisé pour des applications optoélectroniques.