Miniaturization of devices to study chemical interactions and processes in liquid samples has led to the emergence of microfluidics and construction of lab-on-a-chip systems. Present work was devoted to implementation, development and testing of microfluidic systems with detection by confocal Raman microscopy and surface enhanced Raman scattering under the conditions of training department. Several options of performing standard macroscopic measurements in microscopic scales were explored. A method for measuring thermal stability of biopolymers in microsystems with contactless detection of temperature has been designed and tested. Furthermore, possibilites for studying the SERS effect within microfluidic channels were explored. It was demonstrated that the microfluidic chips provide promising opportunity to study hydrodynamics of liquids at microscopic level and chemical reactions and kinetics.

Metabolomics approaches have recently been used to understand the complex molecular interactions of biological systems. One popular area in which these methods are being developed is to understand the biochemical changes during abiotic and biotic stresses; for example, how a cell may respond to a drug. Since metabolites are the end products of gene expression, these can be used to indicate the result of the activities and interaction of the cell or organism with its environment. The investigation of the level and compositional changes of metabolites against metabolic stresses such as chemotherapeutic treatment (drug exposure) are required to understand more fully abiotic perturbation to biological systems. The aim of this project was to understand the metabolic effect that the anti-viral drugs indinavir and lopinavir (currently used by HIV patients) have on HPV-related cervical cancer cell lines by measuring changes in metabolism using a wide range of analytical techniques; including Fourier transform infrared (FT-IR) and Raman spectroscopies, and gas and liquid chromatography-mass spectrometry (GC and LC-MS). The analyses and interpretation of the large volumes of complex multidimensional data generated by metabolomics approaches were performed with a combination of multivariate data analysis techniques such as principal components analysis (PCA) and canonical variates analysis (CVA), as well as univariate approaches such as N-Way analysis of variance (ANOVA). By combining biochemical imaging, metabolite fingerprinting and footprinting, and metabolite profiling, with multi- and uni-variate analyses, the actions and effects of the anti-viral drugs were investigated. FT-IR spectroscopy was initially used to generate global biochemical finger- and foot-prints, and Raman spectroscopy was employed to investigate intracellular distribution of metabolites, and other cellular species, as well as the localisation of drug molecules within cells. FT-IR spectroscopy ascertained that the intra- and extra-cellular metabolomes were being directly influenced in a fashion that correlated with increasing anti-viral dosing; these effects were phenotypic rather than measurements of the drug level. Raman imaging spectroscopy indicated that the indinavir but not lopinavir was being compartmentalised within the cell nucleus, but only in HPV early protein 6 (E6) expressing cells. This observation was further confirmed by fractionation of cell samples into nuclear and cytoplasmic fractions and assessing the indinavir concentrations via LC-MS. Finally, LC-MS and GC-MS metabolite profiling were employed to investigate changes in the intracellular metabolome in response to the anti-viral compounds across a range of physiologically relevant concentrations and in the presence and absence of the E6 oncoprotein. General effects of both anti-viral compounds included the regulation of metabolites such as glutathione, octenedionoic and octadecenoic acids, which may be involved in stress related responses, reduced levels of sugars and sugar-phosphates indicating a potential arrest of glycolysis, and reduced levels of malic acid indicating potential decreased flux into the TCA cycle; all indicating that central metabolism was being reduced. Finally, LC-MS based quantification indicated that in the presence of E6, lopinavir was actively removed from the cell, whereas the indinavir intracellular concentration increased concomitantly with the level of dosing. These investigations have revealed that metabolomics approaches are an apt tool for the study of anti-viral effects within cell cultures, but improvements need to be made with respect to the major limitation of metabolite identification.

Le génome des Influenzavirus A est constitué de huit segments d’ARN de polarité négative formant des ribonucléoprotéines virales (RNPv). La segmentation du génome complique l’empaquetage du génome, puisqu’un jeu complet de segments est nécessaire à l’infectiosité des virus. Il est aujourd’hui admis que le mécanisme d’empaquetage est sélectif et fait intervenir des interactions entre les ARN des différentes RNPv, qui dépendent vraisemblablement de la structure de l’ARN. Notre but a été de caractériser la fixation de la protéine virale NP, composant majeur des RNPv, sur l’ARN in vitro. Nous avons également mis en place une stratégie expérimentale afin de déterminer la structure de l’ARN au sein des particules virales. Ces deux aspects du projet ont été abordés par la cartographie chimique qui permet d’interroger la flexibilité de chaque nucléotide de l’ARN. Nos résultats démontrent une activité chaperone de la protéine NP ainsi que des sites préférentiels de fixation. Notre approche in viro a démontré que l’utilisation de deux sondes chimiques permet de discriminer les interactions ARN-ARN des interactions ARN-NP. / The Influenzavirus genome comprises eight segments of single-stranded RNA of negative polarity packaged in viral ribonucleoproteins (vRNP). Genome segmentation complicates packaging as a full set of vRNPs is required needed for virus infectivity. It is now accepted that packaging is selective and involves interactions between the RNA components of vRNPs that likely depend on the RNA structure. Our goal was to characterize the binding of the viral protein NP, the major component of the vRNP, on in vitro transcribed RNA. We also developed an experimental strategy to determine the RNA structure inside the viral particles. Both parts of the project were addressed with chemical mapping experiments, that interrogates the flexibility of each nucleotide in the RNA structure. Our results show that NP possess an RNA chaperone activity and binds preferential sites. Our in viro approach demonstrate that using two chemical probes allow us to discriminate between RNA-RNA and RNA-NP interactions.

This thesis explores theoretical aspects of scanning transmission electron microscopy (STEM) and the comparison of simulation with experiment. / The long standing contrast mismatch problem between theory and experiment in conventional high resolution transmission electron microscopy (HRTEM) is examined using the principle of reciprocity and bright field scanning transmission electron microscopy (BFSTEM). It is found that quantitative agreement between theoretical and experimental images is possible provided that theory suitably accounts for the spatial incoherence of the source, and that experimental images are placed on an absolute scale with respect to the incident beam current. Agreement between theory and experimental image contrast is found to be independent of specimen thickness and probe defocus. / Core-loss electron energy-loss spectroscopy (EELS) is a powerful experimental tool with the potential to provide atomic-resolution information about the electronic structure at defects and interfaces in materials and nanostructures. Interpretation, however, is nonintuitive due to the nonlocal ionization potential. Novel improvements in microscope design and operating environment have enabled two dimensional chemical maps. This has permitted a more thorough theoretical analysis. This thesis compares experimental STEM EELS images of LaMnO3, BiSrMnO3 and Si samples to the relevant theoretical simulations. Image features which at first appear counter intuitive are discussed and explained with the accompanying theoretical simulations. It is demonstrated, using a sample of SrTiO3, that more direct interpretation of atomic resolution chemical maps is possible when using energy dispersive x-ray spectroscopy (EDS) in STEM. / This thesis considers extending chemical mapping in STEM EELS to three dimensions using depth sectioning. It explores, theoretically, the feasibility to depth section zone-axis aligned crystals that contain embedded impurities. In STEM EELS this is found to be possible for point defects but not for larger extended objects such as nanoparticles. / The theory describing the mechanism by which contrast is obtained in elastic scanning confocal electron microscopy (SCEM) is developed. It is shown that there is no first order phase contrast in SCEM and thus low image contrast. Finally, energy filtered scanning transmission electron microscopy (EFSCEM) is developed theoretically. The fundamental equation describing image formation is derived and an efficient computation method is developed to allow the rapid calculation of EFSCEM images.

La caractérisation fine des matériaux inorganiques nécessite d'avoir accès à des informations complémentaires de celles apportées par des techniques d'analyse élémentaire ou de diffraction. La spectroscopie infrarouge à transformée de Fourier permet de caractériser les liaisons covalentes et l'environnement des groupes fonctionnels dans les matériaux. C'est donc une technique de choix pour l'étude des matériaux hydratés, amorphes ou sujets à des phénomènes de vieillissement. En couplant cette technique à une platine de déplacement, il est possible de réaliser des cartographies des phases sur quelques centimètres carrés : c'est la microscopie infrarouge. Cette thèse développe plus particulièrement l'utilisation de la lumière réfléchie au travers de l'étude de la réflexion spéculaire et de la réflexion totale atténuée (ATR).Après une première partie se focalisant sur les méthodes d'acquisitions disponibles, une seconde partie s'attache à obtenir de manière non supervisée les cartographies chimiques associées aux concentrations relatives des différents composants présents dans la zone analysée. Des techniques de réduction de données et d'analyse factorielle sont mises en place afin d'estimer le nombre de composants chimiques et leurs spectres relatifs ; des problèmes de minimisation sous contraintes sont résolus pour extraire l'information chimique. La réflexion spéculaire ne nécessite aucun contact avec l‘échantillon et, de ce fait, n'entraine aucune altération de la surface analysée. C'est sur le papier une technique de choix pour suivre l'évolution d'un matériau. Cependant, elle souffre de la complexité d'interprétation liée à l'allure des spectres obtenus. Afin de développer la cartographie issue de la réflexion spéculaire, des modèles prenant en compte l'optique géométrique, l'optique ondulatoire, des corrections d'interférogrammes et des méthodes classiques d'homogénéisation ont été développés. Ce travail a permis d'aboutir à un modèle optique liant les spectres issus de la réflexion spéculaire avec les concentrations relatives des composants. Ce modèle tient compte de la polarisation, de l'angle d'incidence et utilise les constantes diélectriques du matériau. Ce modèle a été validé sur un matériau contenant trois composants distincts facilement identifiables en infrarouge et spécialement mis en forme pour cette étude. Ce modèle a ouvert la voie à l'utilisation innovante de lumières polarisées elliptiquement pour déterminer l'indice de réfraction complexe d'un matériau. Ainsi, des spectroscopes infrarouges couplés à un accessoire de contrôle de l'angle d'incidence peuvent être utilisés en complément de l'ellipsométrie / Complementary information, to that provided by elemental analysis and diffraction techniques, is needed to characterize inorganic materials. Fourier Transform Infrared spectroscopy enables to characterize covalent bonds and the environment of functional groups in materials. Thus, it is a technique of interest to study hydrated materials, amorphous materials or any materials, which may experience ageing phenomena. By combining this technique with a micrometric motorized stage, cartographies of chemical compounds can be obtained on several square millimeters: this is the infrared microscopy technique. This Ph.D. thesis focuses on the use of reflected light, in particular through the study of specular reflection and of Attenuated Total Reflectance (ATR). After a first part focused on the different acquisition set-ups, a second part covers the unsupervised methodologies of resolution employed to obtain chemical maps. They result in one map for each component present in the analyzed area. Dimensions reduction techniques and multivariate statistics techniques are implemented to estimate the number of components and their infrared spectra; minimization problems under constraints are solved to retrieve chemical information. When specular reflection is used to acquire spectra, no contact is made with the sample, thus no damage of the analyzed area occurs during the acquisition. A priori, it is a great technique to study the evolution of a material. However, this technique suffers from the complexity of interpretation of the resulting spectra. With the objective to democratize the use of specular reflection to obtain chemical maps, models based on geometrical optics and including diffraction, correction of interferograms and classical homogenization techniques have been developed. This work resulted in an optical model linking the angle of incidence, the polarization state and the dielectric optical constants of the material with the reflected light, which is measured. A model material, constituted of three distinct phases, detectable in the infrared range, has specially been fabricated to validate this optical model. This model set the stage for the use of elliptically polarized light in the determining of the complex refractive indices of materials in the infrared range. Thanks to this development, infrared spectroscopes, equipped with a classical set-up to control the angle of incidence, can now be used in addition to ellipsometry techniques

Cette thèse porte sur l’évaluation des techniques pour la tomographie chimique par STEM EDX : mise au point des procédures expérimentales, traitement des données, reconstruction des volumes, analyse de la qualité des résultats obtenus et évaluation de la complexité globale. Les performances très limitées de l’analyse STEM EDX font que peu d’études, jusqu’à aujourd’hui, se sont portées sur cette technique. Cependant, les avancées très notables procurées par les nouveaux détecteurs ‘SDD’ ainsi que les sources électroniques X-FEG haute brillance, rendant l’analyse STEM EDX 2D très rapide, ont relancé la possibilité de la tomographie chimique ; la technique demande toutefois à être mise au point et évaluée (performances et complexité). Nous avons travaillé sur un microscope Tecnai Osiris permettant d’acquérir des cartographies chimiques EDX de centaines de milliers de pixels avec une résolution de l’ordre du nanomètre en quelques minutes. Nous avons choisi de préparer par FIB des échantillons en forme de pointe et d’utiliser un porte-objet permettant une exploration angulaire de 180° sans ombrage. Puis, à l’aide d’échantillons modèles (billes de SiO2 dans une résine), nous avons évalué les déformations d’échantillon par l’irradiation du faisceau électronique. Ceci nous a permis de proposer une méthode pour limiter cet effet par déposition d’une couche de 20 nm de chrome. Des simulations d’images ont permis d’évaluer les logiciels et méthodes de reconstruction. La méthodologie de chaque étape d’une analyse de tomographie STEM EDX a ensuite été expliquée, et l’intérêt de la technique démontré grâce à la comparaison de l’analyse 2D et 3D d’un transistor FDSOI 28 nm. La qualité des reconstructions (rapport signal-sur-bruit, résolution spatiale) a été évaluée en fonction des paramètres expérimentaux à l’aide de simulations et d’expériences. Une résolution de 4 nm est démontrée grâce à l’analyse d’une mire et d’un transistor « gate all around ». Pour ce même transistor, la possibilité et l’intérêt d’analyse de défaillance à l’échelle nanométrique est prouvée. Une analyse d’un défaut de grille d’une SRAM ou de trous dans un pilier en cuivre permettent d’expliquer l’intérêt d’une combinaison d’un volume HAADF (morphologie et résolution < 4 nm) et du volume EDX (information chimique). La conclusion est que cette technique, qui reste encore à améliorer du point de vue de sa simplicité, montre déjà son utilité pour l’analyse et la mise au point des technologies avancées (nœud 20 nm et après). / This thesis focuses on the evaluation of the STEM EDX chemical tomography technique: development of experimental procedures, data processing and volumes reconstruction, quality analysis of the results and evaluation of the overall complexity. Until now, STEM EDX analysis performances were very limited, so only few studies about this technique have been realized. However, very significant progress procured by the new SDD detectors as well as by the high brightness electronic sources (X-FEG), making the STEM EDX 2D analysis very fast, have revived the possibility of the chemical tomography, although the technique has to be developed and evaluated (performance and complexity). We have worked on a Tecnai Osiris which acquires EDX chemical mapping of hundreds of thousands of pixels with resolution of one nanometer and in a few minutes. We chose to prepare the rod-shaped samples by FIB and use a sample holder allowing an angle of exploration of 180° without shadowing effects. Then, using model samples (SiO2 balls in resin), we evaluated the sample deformation due to the electron beam irradiation. This allowed us to propose a method to reduce this effect by depositing a 20 nm chromium layer. Images simulations were used to evaluate the software and the reconstruction methods. The methodology of each step of the STEM EDX tomography analysis is then explained and the technique interest is demonstrated by comparing the 2D and the 3D analysis of a transistor 28 nm FDSOI. The quality of the reconstructions (signal-to-noise ratio, spatial resolution) was evaluated, in function of experimental parameters, using simulations and experiments. A resolution of 4 nm is demonstrated through the analysis of a test pattern and a "gate all around” transistor. For the same transistor, the possibility and the interest of a failure analysis at the nanoscale is proven. Analyses of a SRAM gate fail or of the holes in a copper pillar explain the benefits of a combination between a HAADF volume (morphology and resolution < 4 nm) and an EDX volume (chemical information). To conclude, this technique, which still needs to be improved in terms of simplicity, is already showing its usefulness for the analysis and the development of advanced technologies (20nm node and beyond).