A simple morphological classification of the galaxies in the local Universe shows two main families: (1) the disc galaxies, with spiral arms and in two-thirds of these galaxies a stellar bar; and (2) the elliptical and lenticular galaxies, labelled early-type galaxies (ETGs), which are dominated by a spheroidal stellar component. ETGs are among the most massive galaxies of the local Universe and present a red color, meaning that their stars are old. These galaxies also present a large diversity of stellar dynamics: they may have a regular rotation pattern aligned with the photometry or perpendicular to it; they can present no global rotation at all; or may hold a central stellar component with a rotation axis distinct from the outer stellar body called a Kinematically Distinct Core (KDC). These features observed in the dynamics of the ETGs and their large mass are clearly signs of past interactions, especially signs of galaxy mergers. The main goal of my thesis is to analyse a large sample of high-resolution numerical simulations of binary galaxy mergers. These binary mergers are called "idealized" because they do not take into account the full cosmological context of galaxy formation: two isolated spiral galaxies are launched in an orbit resulting in a merger of the galaxies, the final remnant is an ETG. The statistical analysis of this large sample of simulations enables us to link the initial conditions of the merger to the final merger remnant. I demonstrated that the mass ratio between the spiral progenitors and the orientation of their spins of angular momentum are the main drivers for the formation of fast and slow rotating ETGs and the KDCs. The morphology of the initial spiral (Bulge/Disc ratio) seems also to play a major role for the formation of the different types of ETGs but its impact is not completelly clear, and other simulations are planned to clarify this problem. During my thesis, I also studied the importance of the resolution in the numerical simulations of galaxy mergers. I showed that the number of particles and the size of the computational grid have a predominant role in the final product of the merger. A too low resolution (i.e. too few particles and a coarse grid) can not follow the rapid evolution of the gravitational potential during the merger. In this case, the angular momentum is not as efficiently transfered to the outer parts of the galaxy: the merger remnant keeps thus a strong and regular rotation. At higher resolution, the scattering of the orbit is resolved and the merger remnant may end-up with, under some special initial conditions, a slow rotation and may form a KDC.

This thesis presents observations and analysis relating to the understanding of processes that govern the formation and evolution of low mass galactic systems. In particular we have focused on separating out the contribution to the chemical and dynamical evolution of dwarf galaxies due to solely secular (internal) processes compared to external effects from the local environment a galaxy resides in. Our observational data focus on an extremely isolated dwarf galaxy, WLM, which we demonstrate has had a uniquely quiescent tidal history, thereby making it an excellent test case for such a study. With spectroscopic and photometric observations of the resolved stars and neutral gas in WLM we have been able to characterize the chemical, structural and kinematic properties of this gas rich dwarf galaxy. As WLM has not been subject to strong tidal or ram-pressure stripping of its stellar and gaseous populations, we have been able to compare the dynamical evolution and chemical history of WLM to theoretical models which are environment independent. A differential comparison of WLM to more environmentally processed dwarf galaxies in the Local Group has revealed that WLM's structural and dynamical state is far from the idealized picture of dIrrs as thin gas-rich rotating systems. The stellar component of WLM shows equal parts rotation and dispersion, and both the gaseous and stellar structural properties show an intrinsically thick axisymmetric configuration. The time evolution of the random (dispersion) component of the stellar orbital energy shows an increase with stellar age, which we show is consistent with secular processes alone - such as disk heating from giant molecular clouds and dark matter substructure. While the degree to which the thick structural and dynamically hot configuration for WLM is surprising, its chemical properties show remarkably consistent values with other galaxies of the same halo mass. Comparing the spatial chemical trends in WLM with other dwarf galaxies we identify a correlation between the strength of the radial abundance gradients and the angular momentum content of dwarf galaxies in the Local Group. Finally using a large sample of chemical abundance measurements in the literature for dwarf galaxies and star clusters, we demonstrate that their distributions of chemical elements all exhibit a binomial form, and use the statistical properties of the distributions to identify a new metric for differentiating low luminosity stellar systems. We further apply a simple binomial chemical evolution model to describe the self-enrichment and pre-enrichment in the two classes of objects, and suggest how this may be used to place constraints on the formation environments of globular clusters in particular. / Graduate

Les galaxies passives présentent des morphologies et propriétés structurelles différentes des galaxies de masse similaire formant des étoiles. La preuve d'une distribution bimodale dans propriétés des galaxies suggère un lien entre les processus de quenching et les structures des galaxies. Contraindre les mécanismes et la chronologie de la formation du bulbe s'avère fondamental pour comprendre l'origine de cette corrélation. Les bulbes grossissent-ils au cours de la séquence principale? Les galaxies ré-accrètent-elles un disque formant des étoiles? Les galaxies stoppent-elles leur formation d'étoile à partir des régions internes? etc. Répondre de manière pertinente à ces questions nécessite de résoudre les parties internes des galaxies à différentes époques. Grâce aux données de haute résolution en multi-longueur d'onde fournies par CANDELS, j'ai réalisé une décomposition séparant le bulbe du disque à partir des courbes de brillance de surface de 17'300 galaxies (F160W<23,0<z<2) dans 4 à 7 filtres couvrant un intervalle spectral compris entre 430 et 1600 nm. Une approche novatrice, basée sur un deep-learning, nous permet de sélectionner a priori les meilleurs profils et de réduire de fait la contamination. J'ai ajusté la SED (densité spectrale d'énergie) avec des modèles de population stellaires (BC03) de disque et de bulbe de manière indépendante afin d'obtenir les paramètres des populations stellaires (masses stellaires, couleurs). Cette procédure fournit un catalogue contenant à la fois les informations structurelles/morphologiques et les propriétés des populations stellaires d'un vaste échantillon de bulbes et de disques galactiques fournit à la communauté (lerma.obspm.fr/huertas/form_CANDELS). Il s'agit du catalogue le plus grand et le plus complet décomposant le bulbe du disque galactique à des redshifts z>0. J'ai utilisé le catalogue ainsi obtenu pour comprendre comment les galaxies stoppent leur formation d'étoile et déterminer l'impact que le quenching peut avoir sur les composantes internes. Les propriétés structurelles des bulbes et des disques, bien que différentes, dépendent peu de la morphologie globale de la galaxie hôte et de son activité de formation d'étoile. Si il existe un seul mécanisme de formation pour tous les types de galaxie ou plusieurs mécanismes contribuant à l'augmentation de la densité centrale, aucune trace dans la structure de la composante interne n'est gardée. De plus, les bulbes et les disques évoluant dans des galaxies soit éteintes, soit formant des étoiles (SF), bien qu'ils présentent des propriétés structurelles similaires, possèdent des distributions de couleurs différentes. Le processus de quenching ne semble pas avoir un impact significatif sur les propriétés des composantes internes.La seconde question clé est de savoir à quel moment les bulbes se forment. La distribution en morphologie le long du graphe SFR-masse montre un manque de galaxie calme (quiescent) avec B/T<0.3 alors que les galaxies avec B/T>0.3 sont présentes tout au long de la séquence principale. Cela suggère que la formation du bulbe doit commencer au cours de la séquence principale. De plus, nous n'avons aucune preuve d'un quelconque processus quenching sans qu'il y ait grossissement du bulbe. Nous n'excluons cependant pas la possibilité que les bulbes de la séquence principale correspondent à des galaxies ayant ré-accrété un disque formant des étoiles. La connaissance des âges est à ce niveau nécessaire pour réellement contraindre ce scénario. Une analyse élargie qui inclurait de l'imagerie à bande étroite (SHARDS) permettrait d'explorer les âges typiques des bulbes et des disques afin de placer des contraintes sur leur temps de formation. / Passive galaxies have different morphologies and structural properties than star-forming galaxies of similar mass. The evidence of a bimodal distribution of galaxy properties suggests a link between the quenching process and and galaxy structure. Understanding the origin of this correlation requires establishing constraints on the mechanisms as well as on the timing of bulge formation. How are bulges formed?Do bulges grow in the main sequence? Are galaxies re-accreting a star forming disk? Do galaxies start to quench from the inside? etc.Proper answers to these questions require resolving the internal components of galaxies at different epochs.Thanks to the CANDELS high-resolution multi-wavelength data, I performed 2-D bulge-disk decompositions of the surface brightness profile of $simeq 17'300$ galaxies (F160W < 23, 0 < z < 2) in 4-7 filters, covering a spectral distribution of 430-1600 nm. A novel approach, based on deep-learning, allowed us to make an a-priori selection of the best profile. Stellar parameters are computed trough the SED fitting. The final catalog contains structural/morphological informations together with the stellar population properties for a large sample of bulges and disks within galaxies. This is the largest and more complete catalog of bulge-disc decompositions at $z>0$.The catalog is then used to investigate how galaxies quench and transform their morphologies.The size of disks and massive bulge is independent of the bulge-to-total ratio ($M_{*}>10^{10} M_{odot}$). It suggests a unique formation process for massive bulges and also that disk survival/regrowth is a common phenomenon after bulge formation. However pure bulges (B/T>0.8), are ~30% larger than bulges embedded in disks at fixed stellar mass and have larger Sersic indices. This is compatible with a later growth of these systems through minor mergers.Bulges in star-forming galaxies are found to be 30% larger than bulges in quenched systems, at fixed stellar mass. Regarding the disks the systematic difference is only a factor of $sim 0.1$. This can be interpreted as a signature that galaxies experience an additional morphological transformation during or after quenching. However, this result is not free of progenitor bias.Moreover, the vast majority (if not all) of pure disks (B/T<0.2) in our sample lie in the main-sequence. It suggests that quenching without any bulge growth is not a common channel at least in the general field environment probed by our data. Pure "blue" bulges (B/T>0.8) do exist however, suggesting that the formation of bulges happens while galaxies are still star forming.Finally, in order to put constraints on the formation times of bulges and disks I analyzed the UVJ colors rest frame. Almost all galaxies in our sample present negative color gradients. Bulges are always redder than the disks at all redshifts. This is compatible with a scenario of inside-out quenching put forward by previous works. However rejuvenation through disk accretion could lead to similar signatures.

Les galaxies massives sont moins nombreuses que ce qui est attendu avec le modèle standard (le modèle Λ-CDM). Ceci ce traduit par une formation d’étoiles moins importante que prévue dans les galaxies. Pour expliquer celà, il est globalement accepté que des processus stoppent le formation d’étoiles. Pour les galaxies massives, ceci est expliqué par l’action des trous noirs supermassifs. En accrétant du gaz, le trou noir central produit de l’énergie et de l’impulsion. Quand l’accrétion devient importante, le trou noir forme un noyau actif de galaxie, et l’énergie peut ralentir la formation d’étoiles, par chauffage du gaz, de la turbulence, ou par ablation du gaz (feedback négatif). Cependant, il existe des cas de feedback positif qui favorise la formation d’étoiles en comprimant le gaz. En particulier, une partie des noyaux actifs produisent des jets de plasma qui sont observés en émission radio. Ces jets radio peuvent intéragir avec du gaz le long de leur direction de propagation. Des telles interactions sont susceptibles de déclencher de la formation d’étoiles (formation induite par les jets). Ma thèse porte sur les interactions jet-gaz dans des radio galaxies proches. J’ai étudié l’effet du jet sur l’efficacité de la formation d’étoiles pour des interactions à des échelles globales (quelques kiloparsecs) et intermédiaires (quelques centaines de parsecs). Pour celà, j’ai observé et cartographié le gaz moléculaire, qui est un élément clé de la formation d’étoiles. Cette phase froide est observable grâce aux équipements au sol actuels de radio astronomie, comme ALMA, APEX, NOEMA et le 30m de l’IRAM. / Massive galaxies are less abundant than predicted by the standard model of galaxy formation (the Λ-CDM model). This means that galaxies form less stars than expected. To explain this behaviour, it is commonly accepted that some processes are at play and quench star formation. For massive galaxies, it is explained by the feedback of the supermassive black holes. While accreting gas, the central black hole produces energy and momentum. When gas accretion becomes important, the black hole forms an active galactic nucleus, and the energy is expected to quench star formation, via gas heating, turbulence or gas removal (negative feedback). However, evidence is found of so-called AGN positive feedback that favours star formation by compressing the gas. In particular, a fraction of the AGN population produces jets of plasma that are observed in radio emission. These radio jets may interact with gas that is located along the direction of propagation. Such interactions are invoked to trigger star formation (jet-induced star formation). My PhD focused on the jet-gas interaction for nearby radio galaxies. I explored the effect of the jet on the star formation efficiency in such interactions at global (few kiloparsecs) and intermediate (few hundreds parsecs) scales. To do so, I searched and mapped the molecular gas (via CO emission lines) that is a key ingredient for star formation. This cold gas is observable using current radio astronomy ground-based facilities, like ALMA, APEX, NOEMA and the 30m telescope.

O estudo de populações estelares em galáxias é particularmente interessante, uma vez que são um registro fóssil de vários processos físicos associados com a formação e evolução das galáxias. Neste trabalho são apresentados os resultados preliminares de uma nova abordagem para o estudo da distribuição de populações estelares no interior das galáxias. A partir das magnitudes ugriz de uma amostra de galáxias, utilizou-se a análise PCA e a determinação de diversos parâmetros para estudar a distribuição espacial das populações estelares nas galáxias. Utilizando uma abordagem píxel a píxel, as populações estelares são investigadas através da aplicação de variadas ferramentas estatísticas, tais como índices de Gini e Funcionais de Minkowski. Esta abordagem é um passo a frente no estudo de galáxias, no sentido que a análise pode ser aprofundada em cada píxel da galáxia, ao invés de estudar a galáxia como um todo. A aplicação de tal tratamento aos píxeis de uma imagem permite a obtenção de estimativas quantitativas sobre a forma como as diferentes populações estelares são distribuídos dentro de uma galáxia, trazendo dicas sobre como elas crescem e evoluem. Nossos resultados preliminares mostraram que uma metodologia píxel a píxel é eficiente no estudo das galáxias. Analisando 15 galáxias de tipos distintos, observou-se que as mesmas possuem populações estelares velhas no centro e novas nas regiões periféricas, corroborando com o cenário inside-out de formação de galáxias. Dada a potencialidade do método desenvolvido, pretende-se automatiza-lo para aplicação em dados de grandes surveys. / The study of stellar populations in galaxies is particularly interesting since they are a fossil record of various physical processes associated with the formation and evolution of galaxies. This work presents the preliminary results of a new approach to the study of the spatial distribution of stellar populations within a galaxy. From the ugriz magnitudes of a sample of galaxies, we used the PCA analysis and determination of various parameters to study the spatial distribution of the stellar populations in galaxies. Using a pixel by pixel approach, the stellar populations are investigated through the application of various statistical tools, such as Gini indexes and Minkowski Functional. This approach is a step forward in the study of galaxies, in the sense that the analysis can be deepened in each pixel of the galaxy, rather than studying the galaxy as a whole. The application of such treatment to the pixels of an image allows to obtain quantitative estimates on how the different stellar populations are distributed within a galaxy, bringing tips on how they grow and evolve. Our preliminary results showed that a pixel by pixel approach is efficient in the study of galaxies. Analyzing 15 galaxies of different types, we observed that they have old stellar populations in the central regions and younger stellar population in the peripheral regions, corroborating the scenario inside-out formation of galaxies. Given the potential of the developed method, the aim is automate and to apply it in data of large surveys.

O estudo de populações estelares em galáxias é particularmente interessante, uma vez que são um registro fóssil de vários processos físicos associados com a formação e evolução das galáxias. Neste trabalho são apresentados os resultados preliminares de uma nova abordagem para o estudo da distribuição de populações estelares no interior das galáxias. A partir das magnitudes ugriz de uma amostra de galáxias, utilizou-se a análise PCA e a determinação de diversos parâmetros para estudar a distribuição espacial das populações estelares nas galáxias. Utilizando uma abordagem píxel a píxel, as populações estelares são investigadas através da aplicação de variadas ferramentas estatísticas, tais como índices de Gini e Funcionais de Minkowski. Esta abordagem é um passo a frente no estudo de galáxias, no sentido que a análise pode ser aprofundada em cada píxel da galáxia, ao invés de estudar a galáxia como um todo. A aplicação de tal tratamento aos píxeis de uma imagem permite a obtenção de estimativas quantitativas sobre a forma como as diferentes populações estelares são distribuídos dentro de uma galáxia, trazendo dicas sobre como elas crescem e evoluem. Nossos resultados preliminares mostraram que uma metodologia píxel a píxel é eficiente no estudo das galáxias. Analisando 15 galáxias de tipos distintos, observou-se que as mesmas possuem populações estelares velhas no centro e novas nas regiões periféricas, corroborando com o cenário inside-out de formação de galáxias. Dada a potencialidade do método desenvolvido, pretende-se automatiza-lo para aplicação em dados de grandes surveys. / The study of stellar populations in galaxies is particularly interesting since they are a fossil record of various physical processes associated with the formation and evolution of galaxies. This work presents the preliminary results of a new approach to the study of the spatial distribution of stellar populations within a galaxy. From the ugriz magnitudes of a sample of galaxies, we used the PCA analysis and determination of various parameters to study the spatial distribution of the stellar populations in galaxies. Using a pixel by pixel approach, the stellar populations are investigated through the application of various statistical tools, such as Gini indexes and Minkowski Functional. This approach is a step forward in the study of galaxies, in the sense that the analysis can be deepened in each pixel of the galaxy, rather than studying the galaxy as a whole. The application of such treatment to the pixels of an image allows to obtain quantitative estimates on how the different stellar populations are distributed within a galaxy, bringing tips on how they grow and evolve. Our preliminary results showed that a pixel by pixel approach is efficient in the study of galaxies. Analyzing 15 galaxies of different types, we observed that they have old stellar populations in the central regions and younger stellar population in the peripheral regions, corroborating the scenario inside-out formation of galaxies. Given the potential of the developed method, the aim is automate and to apply it in data of large surveys.

La première partie de ce manuscrit est consacrée à l'étude de la formation et de l'évolution des galaxies. Pour reconstituer l'histoire de l'évolution des galaxies, le rôle de l'information cinématique est décisif. Dans ce contexte, la spectroscopie intégrale de champ est un outil particulièrement efficace car il permet d'obtenir un diagnostic cinématique complet.<br />Je présenterai dans un premier temps les résultats obtenus grâce à l'instrument GIRAFFE dans le cadre du large programme d'observation IMAGES. Ces résultats montrent que la fraction de galaxies n'ayant pas encore atteint leur équilibre dynamique à z~0.6 est importante. Par ailleurs, l'analyse morphologique détaillée de l'échantillon IMAGES montre que la fraction des galaxies spirales en rotation était environ deux fois plus faible à z~0.6 que dans l'Univers local. En étudiant plus en détails cette population, nous sommes capables pour la première fois d'expliquer la formation d'une partie des galaxies spirales locales.<br />Une des conclusions essentielles de cette étude est qu'en combinant des observations dynamiques et morphologiques, on est désormais en mesure de disséquer et de comprendre l'histoire individuelle de chaque galaxie à z<1. Dans les prochaines années, ces études pourront être étendues à des galaxies à z>>1 grâce à la mise en service des futurs ELTs.<br />Pour cela, ces télescopes géants devront intégrer des systèmes d'Optique Adaptative pour compenser en temps réel la dégradation des images due à la turbulence. La faible luminosité des objets à observer, combinée à la taille des champs à corriger, rend indispensable la mise en place de sources artificielles et d'OA tomographiques. La deuxième partie de cette thèse est donc consacrée à l'étude de l'OA tomographique pour les futurs ELTs.<br />Le travail présenté porte en particulier sur l'étude et l'analyse des limites fondamentales des techniques tomographiques. En nous appuyant sur des développements théoriques dans une base de Fourier, nous proposons plusieurs voies pour l'optimisation de ces futurs systèmes.<br />La dernière partie de cette thèse est consacrée à l'étude de EAGLE: un projet de spectrographe intégrale de champ multi-objets pour l'ELT européen.<br />A l'aide des résultats obtenus avec GIRAFFE pour la définition des critères scientifiques et de l'analyse des techniques tomographiques, nous proposons un premier design d'optique adaptative capable de répondre aux besoins de la spectroscopie intégrale de champ des galaxies à z>>1. <br />En alliant la haute résolution spatiale et spectrale avec le pouvoir collecteur immense des futurs ELTs, l'instrument EAGLE apportera certainement des indices décisifs dans notre compréhension des processus à l'oeuvre pour l'assemblage de la masse dans les galaxies.

Les observations montrent qu'il y a dix milliards d'années, les galaxies formaient bien plus d'étoiles qu'aujourd'hui. Comme les étoiles se forment à partir de gaz moléculaire froid, cela signifie que les galaxies disposaient alors d'importants réservoirs de gaz, et c'est ce qui est observé. Mais les processus de formation d'étoiles pourraient aussi avoir été plus efficaces : qu'en est-il ? Les étoiles se forment dans des nuages moléculaires géants liés par leur propre gravité, mais les toutes premières étapes de leur formation demeurent relativement mal connues. Les nuages moléculaires sont eux-mêmes fragmentés en différentes structures, et certains scénarios suggèrent que les filaments interstellaires qui y sont observés aient pu constituer la première étape de la formation des coeurs denses dans lesquels se forment les étoiles. En quelle mesure leur géométrie filamentaire affecte-t-elle les coeurs pré-stellaires ? Des phenomènes de rétroaction liés à l'évolution des étoiles, comme les vents stellaires et les explosions de supernovae, participent à la régulation de la formation d'étoiles et peuvent aussi perturber la distribution de matière noire supposée entourer les galaxies. Cette thèse aborde l'évolution des galaxies et la formation des étoiles suivant trois perspectives : (i) la caractérisation des processus de formation d'étoiles à des échelles sous-galactiques au moment de leur pic de formation ; (ii) la formation des coeurs pré-stellaires dans les structures filamentaires du milieu interstellaire ; et (iii) les effets rétroactifs de la formation et de l'évolution des étoiles sur la distribution de matière noire des galaxies. / Observations show that ten billion years ago, galaxies formed their stars at rates up to twenty times higher than now. As stars are formed from cold molecular gas, a high star formation rate means a significant gas supply, and galaxies near the peak epoch of star formation are indeed much more gas-rich than nearby galaxies. Is the decline of the star formation rate mostly driven by the diminishing cold gas reservoir, or are the star formation processes also qualitatively different earlier in the history of the Universe? Ten billion years ago, young galaxies were clumpy and prone to violent gravitational instabilities, which may have contributed to their high star formation rate. Stars indeed form within giant, gravitationally-bound molecular clouds. But the earliest phases of star formation are still poorly understood. Some scenarii suggest the importance of interstellar filamentary structures as a first step towards core and star formation. How would their filamentary geometry affect pre-stellar cores? Feedback mechanisms related to stellar evolution also play an important role in regulating star formation, for example through powerful stellar winds and supernovae explosions which expel some of the gas and can even disturb the dark matter distribution in which each galaxy is assumed to be embedded. This PhD work focuses on three perspectives: (i) star formation near the peak epoch of star formation as seen from observations at sub-galactic scales; (ii) the formation of pre-stellar cores within the filamentary structures of the interstellar medium; and (iii) the effect of feedback processes resulting from star formation and evolution on the dark matter distribution.

This thesis presents spectroscopic observations of z ≥ 7 galaxy candidates in the Hubble Ultra Deep Field, which were selected with HST WFC3 imaging, using the Lyman-Break technique. Four z-band (z ≈ 7) dropout galaxies were targeted with Gemini/GNIRS, one z-band dropout galaxy and three Y -band (z ≈ 8 − 9) dropout galaxies with VLT/XSHOOTER, and 22 z-band dropouts with VLT/FORS2, where 15 of the latter are strong candidates. No evidence of Lyman-α emission is found, and the upper limits on the Lyman-α flux and the broad-band magnitudes are used to constrain the rest-frame equivalent widths for this line emission. Amongst the targeted objects, observations were made of HUDF.YD3, a relatively bright Y -band dropout galaxy likely to be at z ≈ 8 − 9 on the basis of its colours in the HST ACS and WFC3 images. Lehnert et al. (2010) observed this galaxy using the VLT/SINFONI integral field spectrograph and claim that it exhibits Lyman-α emission at z = 8.55. In observations of this object described in this thesis, which were made with VLT/XSHOOTER and Subaru/MOIRCS, this line was not reproduced despite the expected signal in the combined MOIRCS & XSHOOTER data being 5σ. Hence it appears unlikely that the reported Lyman-α line emission at z > 8 is real. Accounting for incomplete spectral coverage, in total (across all spectro- graphs) 9.63 z-band dropouts and 1.15 Y -band dropouts are surveyed to a Lyman-α rest-frame Equivalent Width better than 75 ̊A. A model where the fraction of high rest-frame equivalent width emitters follows the trend seen at z = 3−6.5 is inconsistent with these non-detections at z = 7−9 at a confidence level of ∼ 91%, which may indicate that a significant neutral HI fraction (χHI) in the intergalactic medium suppresses the Lyman-α line at z > 7. In particular, the lack of detection of Lyman-α emission in this spectroscopy is compared with results at lower redshift by Stark et al. (2010), who derive a mapping between Lyman-α fractions and χHI based on radiative transfer simulations by McQuinn et al. (2007). These results suggest a lower limit of χHI ~ 0.5.

L'etude des raies d'absorption dans le spectre des quasars constitue<br />aujourd'hui un des outils majeurs de l'observation des galaxies au<br />cours de leur formation et de leur evolution. En effet, ces raies<br />revelent le gaz interpose tout au long de la ligne de visee entre le<br />quasar et l'observateur. Les systemes avec les plus grandes densites<br />de colonne, les sytemes lorentziens (DLAs), ont une densite de colonne<br />en hydrogene neutre superieure a 2 * 10^20 atomes cm^-2. Ces<br />absorbants ont longtemps ete consideres comme les precurseurs des<br />galaxies spirales observees a faible decalage spectral (z), meme si<br />certains travaux semblent indiquer qu'ils pourraient etre constitues<br />en majorite de galaxies naines riches en gaz, elements de base de la<br />formation hierarchique des structures. L'observation de ces systemes<br />etant independante de leur taille, forme ou facteur de recouvrement,<br />ils fournissent des informations complementaires a l'etude de<br />l'emission propre des galaxies. Les systemes Lyman limit (LLSs)<br />possedent une densite de colonne N(HI) > 1.6 * 10^17 atomes cm^-2. A<br />z<1, ils sont probablement associes aux halos des galaxies. Enfin, les<br />nombreux systemes avec les plus petites densites de colonne (10^12 -<br />10^17 atomes cm^-2) constituent "la foret Lyman-alpha".<br /><br /><br />Cette these presente un echantillon de 66 quasars brillants a z>4,<br />observes avec le telescope de 4 m "Cerro Tololo Inter-American" et<br />le telescope de 4,2 m "William Herschel". En premier lieu, l'etude<br />consiste en l'analyse des quasars a l'aide de la modelisation de<br />l'emission continue du quasar, et de la mesure des parametres de<br />depression du continu qui quantifient l'absorption moyenne dans la<br />region de la foret Lyman-alpha. Les spectres de quasars sont ensuite<br />analyses de maniere a etudier les absorbants qu'ils contiennent. Ces<br />donnees menent a la construction d'un echantillon de 26 nouveaux<br />systemes lorentziens lointains et 34 nouveaux systemes Lyman limit<br />ainsi que de nombreux systemes metalliques associes. Cet ensemble de<br />donnees permettent de determiner la distribution de la densite de<br />colonne, f(N,z), ainsi que la masse totale d'hydrogene neutre contenue<br />dans les absorbants de grande densite de colonne. Le nombre de LLSs<br />observes par unite de decalage spectral permet de contraindre f(N,z)<br />en-dessous de la limite des DLAs, dans l'intervalle N(HI) = 1,6 *<br />10^17 a 2 * 10^20 atomes cm^-2. L'analyse demontre sans ambiguite que<br />f(N,z) diverge d'une loi de puissance et qu'une distribution gamma<br />f(N,z) = (f_*/N_*)(N/N_*)^(-beta) exp(-N/N_*) represente de maniere<br />plus precise les observations. Ces resultats sont ensuite utilises<br />pour determiner la quantite de gaz neutre contenue dans les systemes<br />DLAs ainsi que dans les absorbants avec N(HI) > 2 * 10^19 atoms cm^-2<br />(systemes "sous-lorentziens"). Dans l'intervalle de decalage<br />spectral 2 - 3, 85% de la densite en masse HI + He II se trouve dans<br />les DLAs. Cependant, a z>3,5, nous trouvons que cette fraction est<br />seulement de 55%, le reste de la masse se trouvant dans les systemes<br />sous-lorentziens. Apres avoir corrige la masse HI observee en prenant<br />en compte ce gaz neutre "manquant", la densite comobile en masse ne<br />presente plus de decroissance dans l'intervalle z = 2 - 5 par rapport<br />aux etudes anterieures ne considerant pas les systemes<br />sous-lorentziens. Le changement dans la distribution de densite de<br />colonne suggere qu'a z>3,5, nous observons directement la formation<br />des systemes de grande densite de colonne HI a partir de systemes de<br />densite de colonne plus petites. Enfin, nous presentons nos<br />predictions concernant l'evolution de la densite en nombre de systemes<br />sous-lorentziens avec le decalage spectral. Des resultats<br />preliminaires de cette mesure a partir de donnees d'archives echelle<br />UVES semblent etre en accord avec ces predictions.