Four regions of massive star formation in the Large Magellanic Cloud (LMC) were observed for water and methanol maser emission and radio continuum emission. A total of 42 radio detections were made including 27 new radio sources, four water masers, and eight compact HII regions. The lobes of a radio galaxy were resolved for the first time, and the host galaxy identified. Seven sources were associated with known massive young stellar objects (YSOs). A multi-wavelength analysis using both the infrared and radio spectrum was used to characterize the sources. Mid-infrared color-magnitude selection criteria for ultracompact HII (UCHII) regions in the LMC are presented, yielding 136 UCHII region candidates throughout that galaxy. New maser detections identified two previously unknown massive YSOs. No methanol masers were detected, consistent with previous studies and supporting the hypothesis that the LMC may be deficient in these molecules. These discoveries contribute to the history of star formation in the LMC, which will lead to a better understanding of star formation in the Milky Way and throughout the universe.

The mechanism by which massive stars form is not nearly as well understood as it is for lower mass stars. For instance, at the onset of massive star formation, it is still not clear whether the mass for a given massive star comes from the turbulent collapse of a dense core (i.e McKee & Tan, 2003) or whether the star continues to accrete material from the cores environment as it grows (i.e. Bonnell et al., 1998). From this point, it is suggested that the cold, massive core (an Infrared Dar Cloud) begins to heat up and form a Hot Core. Later in its protostellar evolution, an HII region forms from the ionizing radiation being produced by the massive star. How, or even whether, accretion onto the massive protostar can continue in the presence of the large outward thermal and radiation pressures from the star is also quite uncertain. Can the star continue to accrete ionized gas (i.e. Keto & Wood, 2006)? Are the accretion rates high enough early on to account for the final observed masses (i.e. Klaassen et al., 2006)? Or, is there some way of minimizing the radiation pressure affecting the infalling gas (i.e. McKee & Ostriker, 2007, and references therein). Here, we present observations which suggest that there is a statistically significant, although short, period in which rotation and infall of molecular gas (which powers a bipolar outflow) continue after the formation of an HII region. This continued infall of material is seen on both large and small scales, and appears to be continuing to produce outflows in many of the sources observed in this study. That it is not seen in all sources suggests that this stage is short lived. / Thesis / Doctor of Philosophy (PhD)

The majority of stars form in clusters which are themselves birthed in Giant Molecular Clouds (GMCs). The radiation produced by clusters during their formative phase heats and ionizes the surrounding gas and drives outflows via radiation pressure. The combination of these processes, referred to as radiative feedback, is a proposed mechanism for limiting the star formation efficiency (SFE) in molecular clouds. In this thesis, we use 3D numerical simulations of turbulent GMCs, completed using the code FLASH and a sophisticated radiative transfer scheme, to explore the effects of radiative feedback on cluster formation and the larger scale interstellar medium (ISM). We present suites of simulations that vary the initial GMC mass from 10^4 to 10^6 M$_{\odot}$ and consider both gravitationally bound and unbound clouds. We find that clusters form within the highly filamentary clouds where they can undergo subsequent merging. Radiative feedback only plays a minor role in lowering the SFE of 10^6 M$_{\odot}$ GMCs. However, it completely disrupts intermediate mass clouds (~10^5 M$_{\odot}$), reducing the SFE by a factor of two. We then examine the escape fraction of UV photons from GMCs --- a quantity relevant to the structure of the ISM and cosmic reionization. We show that the escape fraction is dynamic and can vary by factors of two over short timescales because of the rapid growth and collapse of HII regions. The escape fractions from massive GMCs are typically low (~5%) while intermediate mass models are characterized by escape fractions nearing 100%. We combine our GMC models to represent the escape fraction from a population of clouds in dwarf starburst and spiral-type galaxies. We successfully reproduce the star formation rates in these galaxies and find typical escape fractions of 8% in all cases. These results place important constraints on galactic-scale models studying the ISM and cosmic reionization. / Thesis / Doctor of Philosophy (PhD)

Nesta tese é apresentado um catálogo fotométrico no infravermelho próximo de 35 Regiões HII, todas pertencentes ao disco Galáctico. Esta faixa espectral é útil uma vez que os comprimentos de onda são grandes o suficiente para se ter uma baixa extinção interestelar comparada ao visível, e são pequenos o suficiente para diagnosticar as fotosferas estelares. Foram obtidas imagens nas bandas J, H e K e imagens do Spitzer nos canais de 3,5, 5,8 e 8,0 m. Após a fotometria nas imagens JHK, foi possível construir diagramas cor-cor e cor-magnitude. Foram utilizadas imagens coloridas, compostas de uma combinação RGB das imagens nas três bandas, tanto para as imagens JHK quanto para as imagens do Spitzer. Estas imagens, junto com os diagramas, foram utilizadas para levantar candidatos a fontes ionizantes das regiões HII, assim como objetos ainda em estágios primordiais de evolução (CTTs e MYSOs). Estes dados também foram utilizados para associar à cada região HII um estágio evolutivo (de A até D, da região mais jovem à mais evoluída). Baseado na posição da Sequência Principal em diagramas cor-magnitude, foi possível comparar as distâncias cinemáticas com nossos dados. Além disso, quando possível, foram utilizadas distâncias de regiões HII determinadas por paralaxe espectrofotométrica (disponíveis na literatura) e utilizando duas leis de extinção interestelar extremas mostrou-se que estas distâncias são menores que suas contrapartidas cinemáticas, e estão em acordo com distâncias determinadas por outros métodos, como por paralaxe trigonométrica. Sabendo que estas regiões de formação estelar seguem a dinâmica do gás, o mapeamento da distribuição destas regiões permite checar a estrutura espiral da Via Láctea. / In this work, a near infrared photometric catalog of 35 HII regions that belongs to the Galactic plane is presented. This spectral range is useful since the wavelengths are long enough to have less influence of the interstellar extinction compared to the visible domain, and they are small enough to still show stellar photospheric features. Images of these HII regions in the J, H and K-band together with IRAC-Spitzer images (channels 4.5, 5.8 and 8.0 m) were used. After the photometry in the JHK images, color-color and color-magnitude diagrams were constructed. These two group of images (JHK and 4.5, 5.8 and 8.0 m) colored in a RGB combination were used, together with the diagrams, to identify the ionizing sources candidates, as well as objects still embedded in their natal cocoon (CTTs and MYSOs). An evolutionary stage to these regions (from A to D, from the younger region to the more evolved) was inferred based on the images and diagrams. These diagrams were also used to infer if the kinematic distance is correct, based on the Main Sequence location. Non-kinematic distances to several HII regions, when it was possible, were collected from the literature. Using two extreme interstellar extinction laws, it was possible to compare these distances with the kinematic results. These non-kinematic distances are lower than that from kinematic techniques. Also, these distance discrepancies are in agreement with distances derived by others methodologies, as trigonometric parallax. Since these star forming regions follow the gas dynamics, mapping their distribution along the Galaxy allows to check the spiral pattern of the Milky Way.

Conselho Nacional de Desenvolvimento Científico e Tecnológico / Quantitative measurements of line intensities have been made since the beginning of the last century. The quantity and quality of the data available have increased drastically with the advent of electronic detectors. However, the data are widely scattered in the literature. In face of this, we compiled a catalog of emission-lines intensities. It contains data on HII regions in 41 galaxies, all totaling approximately 36 000 line intensities. We have constructed a table with 1 750 emission-lines observed, showing their transition probabilities and configurations or terms. We have standardized the identification of all the emission-lines of the catalog based on this table. The database was used for different studies. We checked the values of the ratios [OIII] λ5007/λ4959 and [NII] λ6583/λ6548 with the emission-line ratios from the HII regions in the catalog. The values found are in agreement with the theoretical ones. With the diagram λ6717/Hβ versus λ6731/Hβ, we verified that most of the objets in the catalog are near the low-density limit. We also present a comparison of different temperature indicators based on observational data from HII regions, HII galaxies and planetary nebulae extracted from the literature. We have found that there is not a unique relation between the temperature indicators of different ionization zones which allows a reliable derivation of one of these from another. We have constructed diagnostic diagrams comparing the emission-lines ratios [OI]/Hα, [OII]/Hβ, [OIII]/Hβ, [NII]/Hα, [SII]/Hα e [OII]/[OIII]. For them, we have distinguished the emission-line ratios of HII regions of different galaxies and, for data from our galaxy, we have distinguished different objects. We described the diagnostic diagrams using theoretical grids varying the ionization parameter and the chemical abundance. These were computed using a photoionization code and different spectral energy distributions. With the diagram [NII]/[OII] versus [OIII]/[OII], we have obtained the best separation between ionization parameter and metallicities. We observed large discrepancies between the chemical abundances of the grids of models and those determined from collisionaly excited lines. / Medidas quantitativas de intensidades de linhas de emissão têm sido feitas desde o in´ıcio do século passado. A quantidade e a qualidade dos dados disponíveis têm crescido drasticamente com o advento dos detectores eletrônicos. Porém, os dados estão amplamente espalhados na literatura. Diante disso, compilamos um catálogo de intensidades de linhas de emissão. Este contém dados de regiões HII de 41 galáxias, totalizando aproximadamente 36 000 intensidades de linhas. Construímos uma tabela com 1 750 linhas de emissão observadas, apresentando suas probabilidades de transição e configurações ou termos. Padronizamos a identificação de todas as linhas de emissão do catálogo com base nessa tabela. O banco de dados foi utilizado para diferentes estudos. Checamos os valores das razões [OIII] λ5007/λ4959 e [NII] λ6583/λ6548 com razões de linhas de emissão de regiões HII do catálogo. Os valores encontrados concordam com os teóricos. Com o diagrama λ6717/Hβ versus λ6731/Hβ, verificamos que a maior parte dos objetos do catálogo estão próximos ao limite de baixa densidade. Apresentamos também uma comparação de diferentes indicadores de temperatura baseada em dados observacionais de regiões HII, de galáxias HII e de nebulosas planetárias extraídos da literatura. Encontramos que não há uma única relação entre indicadores de temperaturas de diferentes zonas de ionização que permita uma derivação confiável de uma destas a partir da outra. Construímos diagramas de diagnóstico comparando as razões de linhas [OI]/Hα, [OII]/Hβ, [OIII]/Hβ, [NII]/Hα, [SII]/Hα e [OII]/[OIII]. Para eles, distinguimos as razões de linhas de emissão de regiões HII de galáxias diferentes e, para os dados de nossa galáxia, distinguimos os diferentes objetos. Descrevemos os diagramas de diagnóstico utilizando grades teóricas variando o parâmetro de ionização e a metalicidade. Estas foram calculadas usando um código de fotoionização e diferentes distribuições espectrais de energia. Com o diagrama [NII]/[OII] versus [OIII]/[OII], obtivemos a melhor separação entre o parâmetro de ionização e a metalicidade. Verificamos grandes discrep ancias entre as abundâncias químicas das grades de modelos e aquelas determinadas de linhas excitadas colisionalmente.

Nesta tese é apresentado um catálogo fotométrico no infravermelho próximo de 35 Regiões HII, todas pertencentes ao disco Galáctico. Esta faixa espectral é útil uma vez que os comprimentos de onda são grandes o suficiente para se ter uma baixa extinção interestelar comparada ao visível, e são pequenos o suficiente para diagnosticar as fotosferas estelares. Foram obtidas imagens nas bandas J, H e K e imagens do Spitzer nos canais de 3,5, 5,8 e 8,0 m. Após a fotometria nas imagens JHK, foi possível construir diagramas cor-cor e cor-magnitude. Foram utilizadas imagens coloridas, compostas de uma combinação RGB das imagens nas três bandas, tanto para as imagens JHK quanto para as imagens do Spitzer. Estas imagens, junto com os diagramas, foram utilizadas para levantar candidatos a fontes ionizantes das regiões HII, assim como objetos ainda em estágios primordiais de evolução (CTTs e MYSOs). Estes dados também foram utilizados para associar à cada região HII um estágio evolutivo (de A até D, da região mais jovem à mais evoluída). Baseado na posição da Sequência Principal em diagramas cor-magnitude, foi possível comparar as distâncias cinemáticas com nossos dados. Além disso, quando possível, foram utilizadas distâncias de regiões HII determinadas por paralaxe espectrofotométrica (disponíveis na literatura) e utilizando duas leis de extinção interestelar extremas mostrou-se que estas distâncias são menores que suas contrapartidas cinemáticas, e estão em acordo com distâncias determinadas por outros métodos, como por paralaxe trigonométrica. Sabendo que estas regiões de formação estelar seguem a dinâmica do gás, o mapeamento da distribuição destas regiões permite checar a estrutura espiral da Via Láctea. / In this work, a near infrared photometric catalog of 35 HII regions that belongs to the Galactic plane is presented. This spectral range is useful since the wavelengths are long enough to have less influence of the interstellar extinction compared to the visible domain, and they are small enough to still show stellar photospheric features. Images of these HII regions in the J, H and K-band together with IRAC-Spitzer images (channels 4.5, 5.8 and 8.0 m) were used. After the photometry in the JHK images, color-color and color-magnitude diagrams were constructed. These two group of images (JHK and 4.5, 5.8 and 8.0 m) colored in a RGB combination were used, together with the diagrams, to identify the ionizing sources candidates, as well as objects still embedded in their natal cocoon (CTTs and MYSOs). An evolutionary stage to these regions (from A to D, from the younger region to the more evolved) was inferred based on the images and diagrams. These diagrams were also used to infer if the kinematic distance is correct, based on the Main Sequence location. Non-kinematic distances to several HII regions, when it was possible, were collected from the literature. Using two extreme interstellar extinction laws, it was possible to compare these distances with the kinematic results. These non-kinematic distances are lower than that from kinematic techniques. Also, these distance discrepancies are in agreement with distances derived by others methodologies, as trigonometric parallax. Since these star forming regions follow the gas dynamics, mapping their distribution along the Galaxy allows to check the spiral pattern of the Milky Way.

Les propriétés du milieu interstellaire (MIS) influencent fortement l’environnement et les processus menant à la formation d’étoiles qui, à son tour, dicte l’évolution d’une galaxie. Les galaxies naines du Groupe Local sont de parfaits laboratoires pour comprendre comment le contenu en métaux (ou métallicité) du MIS affecte l’interaction entre le gaz, la poussière et les étoiles. Mon travail de thèse porte sur les propriétés physiques des régions HII et du gaz diffus ionisé de la galaxie naine IC10, de métallicité 1/3 solaire. La proximité de cette galaxie (d=700kpc) permet l’analyse du MIS à différentes échelles spatiales: des nuages brillants compacts (25pc) au corps entier de la galaxie formant des étoiles (650pc). Afin de mesurer les propriétés physiques du MIS, j’ai modélisé les raies d’émission en infrarouge observées avec Spitzer et Herschel grâce à des modèles de photoionisation et de photodissociation. Je présente une exploration complète de plusieurs méthodes pour déterminer, de manière la plus fiable et selon les contraintes disponibles, les propriétés du MIS à diverses échelles. J’ai contraint les propriétés des nuages compacts les plus brillants dans IC10 et montré que l’émission à plus grande échelle (300pc) est dominée par celle de ces nuages. Enfin, je démontre le besoin d’un modèle à plusieurs composantes pour reproduire les observations dans leur ensemble. / The properties of the Interstellar Medium (ISM) strongly influence the environment and processes that lead to star-formation, which in turn, drives the evolution of a galaxy. Dwarf galaxies in the Local Group are perfect laboratories to investigate how the metal-poor ISM affects the interplay between gas, dust and stars. In this thesis, I investigate the properties of the HII regions and the diffuse ionized gas of the nearby dwarf galaxy IC10, which has a metallicity of 1/3 solar. Its proximity (d=700 kpc) enables the analysis on different spatial scales: from the compact clumps (~25 pc) to the whole star-forming body of the galaxy (~650pc). In order to measure the physical properties of the ISM, I model the infrared emission lines observed with Spitzer and Herschel with photoionization and photodissociation models. I present an extensive exploration of different methods to determine the most reliable ISM properties, based on the available constraints. I determined the properties of the brightest star-forming clumps within the galaxy and show that the emission at large scales (~300 pc) is dominated by that of the compact, bright clumps that lie within the region. I further demonstrate the need for a multi-component model to fully reproduce the observations.

Conselho Nacional de Desenvolvimento Científico e Tecnológico / In order to determine distances of Southern Galactic HII regions, we carried out spectroscopic and photometric observations of the stellar content of a set of these objects. We have chosen objects only visible in the South hemisphere, with no published or uncertain distances. In this work, we present spectral classification based on intermediate dispersion spectra with a very high signal to noise ratio and on stellar UBV photometry, so that the distances have been determined using the usual method, the spectroscopic parallax. We picked up the 2MASS infrared photometry and assessed the reddening individually for each star, using the color-difference method. This way, we have estimated the totalto-selective extinction ratio for the most of the classified stars, and so, we have found a mean value 21% higher than the mean value of RV calculated on the whole directions of the Galaxy. As a result, we have found distances in general smaller than those already published on the literature, although confirming several spectral classifications. Besides that, large fluctuations around the values normally adopted were also observed, which would result in larger discrepancies between the stellar distances. At last, we have also extracted nebulae spectra in order to estimate the kinematics distances, so that we could compare the distance measurement results by different methods. In general, we have found reasonably compatible distances. However, in some cases large differences was found, suggesting that some objects present deviations of the circular motion. / Com o objetivo de determinar distâncias de regiões HII Galácticas observáveis no Hemisfério Sul, realizamos observações espectroscópicas e fotométricas do conteúdo estelar de um conjunto desses objetos. Demos preferência para aqueles objetos cuja observação só é possível no Hemisfério Sul, sem publicações de distância encontrada ou com grande dispersão entre os valores publicados. Neste trabalho, apresentamos classificação espectral baseada em espectros de dispersão intermediária, com elevada razão sinal / ruído e, também, classificação espectral via fotometria UBV , ambas com o objetivo de determinar as distâncias das regiões HII via paralaxe espectroscópica de suas estrelas ionizantes. Nós também extraímos as magnitudes dessas estrelas no infravermelho próximo diretamente do catálogo 2MASS e calculamos o avermelhamento individualmente para cada estrela usando o método das diferenças de cores. Assim, a razão entre a extinção total e a seletiva a banda V (RV ) foi estimada para a maior parte das estrelas, de forma que encontramos um valor médio 21% maior que o valor médio de RV calculado sobre todas as direções da Galáxia e, por este motivo, nós encontramos distâncias heliocêntricas, em geral, menores do que aquelas encontradas na literatura, mesmo confirmando a maior parte das classificações espectrais dadas por esses autores. Além disso, grandes flutuações em torno do valor médio foram encontradas, o que acarretaria maiores discrepâncias entre as distâncias estelares. Por fim, extraímos espectros nebulares dos objetos, com o objetivo de estimar suas distâncias cinemáticas e comparar os resultados obtidos com diferentes métodos. Em geral, encontramos distâncias razoavelmente compatíveis. Contudo, em alguns casos, grandes diferenças foram encontradas, sugerindo que alguns objetos como RCW 88 e NGC 3503 apresentam desvios do movimento circular.

In recent years the 6.67 GHz masing transition of CH3OH has proven to be a superior tracer of massive star formation (see Minier 2001). Maser sites often occur in proximity to UCHII regions, however, up to 75 per cent of sites have no detectable radio counterpart (Walsh 1998) and are instead hypothesised to trace the less evolved 'hot molecular core' phase of stellar evolution. This has been confirmed for a only handful of well known sources (e.g., Cesaroni 1994). Presented here are the results of multi-species molecular line observations towards warm, dusty clumps, undertaken with the goal of investigating the relationship between hot cores, UCHII regions and CH3OH masers. Data from the 22-m Mopra telescope is used extensively in this thesis and substantial efforts were made to calibrate the brightness temperature scale. Measurements conducted on SiO masers and planets show that the beam pattern is divided into a Gaussian main beam plus an inner error lobe, which in 2004 contained 1/3 of the power in the main beam. Full-width half-maximum beam sizes were measured from the data and the beam efficiencies were derived for the years 2000-2004. A 3-mm wavelength molecular line survey was conducted, using Mopra, towards 83 massive star-forming clumps associated with CH3OH masers. Emission from the transitions 13CO (1-0), N2H+ (1-0), HCO+ (1-0), HCN (1-0) and HNC (1-0) was detected towards 82 sources (99 per cent), while CH3OH emission was detected towards 78 sources (94 per cent). The warm gas tracer CH3CN was observed specifically to search for hot core chemistry, and was detected towards 58 sources (70 per cent), confirming that CH3OH masers are excellent tracers of hot cores. CH3CN is found to be brighter and more commonly detected towards masers associated with UCHII regions compared to 'isolated' masers. That CH3CN is detected towards isolated maser sources strongly suggests that these objects are internally heated. The molecular line data have been used to derive rotational temperatures and chemical abundances in the clumps and these properties have been compared between sub-samples associated with different indicators of evolution. In particular, CH3OH is found to be brighter and more abundant in UCHII regions and in sources with detected CH3CN, and may constitute a crude molecular clock in single dish observations. Gas-kinematics were analysed via asymmetries in the HCO+ line profiles. Approximately equal numbers of red and blue-skewed profiles, indicative of inward or outward motions, respectively, are found among all classes of object. Bolometric luminosities were derived via greybody fits to the sub-millimetre and mid-infrared spectral energy distributions, and an empirical gas-mass to luminosity relation of L proportional to M^0.68 was fit to the sample. This is a considerably shallower power law than L proportional to M^3 for massive main-sequence stars. In the mid-infrared, 12 sources were identified as 'infrared dark clouds' (IRDCs). Such objects have been hypothesised as precursors to the hot core phase of evolution, however, we find these sources have greater linewidths and rotational temperatures than the bulk of the sample, and one contains an embedded HII region The filamentary star forming region NGC3576 was also investigated via a molecular line and 23 GHz continuum mapping survey, utilising the ATCA, Mopra and Tidbinbilla telescopes. The results of these observations provide detailed information on the morphology, masses, kinematics, and physical and chemical conditions along the cloud. Analysis of NH3 data has revealed that the temperature and linewidth gradients exist in the western arm of the filament. Values are highest near to the central HII region, indicating that the embedded cluster of young stars is influencing the conditions in the bulk of the gas. Six new H2O masers were detected in the arms of the filament, all associated with clumps of NH3 emission. Star formation is clearly underway, however, clump masses range from 1 to 128 solar masses, possibly too low to harbour very massive stars. The lack of detected 23 GHz continuum emission in the arms supports this assertion.

La raie hyperfine à 21-cm de l'hydrogène atomique émise dans l'univers à grand redshift (6 < z < 20) permettra de sonder une nouvellère de la cosmologie, l'époque de la réionisation (EoR). Elle fournira des informations plus détaillées, moins ambigües et plus complètes en 3D que les autres observations possibles de l'EoR (comme les raies d'absorption des quasars, ou le CMB). Son intensité dépend de nombreux processus physiques. Les radiotélescopes de nouvelle génération, SKA et ses précurseurs, entreront en service dans les prochaines années pour observer ce signal. La simulation numérique du 21-cm est importante pour optimiser le design des instruments et interpréter les observations futures. <br />Dans ce travail, nous avons développé un module de transfert radiatif pour le continuum ionisant dans le code LICORICE dans le but de simuler l'époque de la réionisation, pour laquelle le transfert radiatif est un outil essentiel. Nous utilisons un algorithme de ray-tracing de type Monte Carlo sur une grille adaptative. Plusieurs tests sont réalisés pour valider le code dans deux cas, celui d'un champ de densité statique et dans le cas de l'hydrodynamique radiative. <br />Nous produisons ensuite le signal à 21-cm émis pendant l'EoR, qui nous fournit une sonde directe de la réionisation et contient de nombreuses informations sur les sources d'ionisation et de chauffage. Habituellement, on suppose que le flux Ly-α, qui influence la puissance du signal à 21-cm, est fort et homogéne pendant toute la réionisation, mais cette hypothèse n'est pas valable durant la première phase de l'EoR. Nous évaluons exactement le flux Ly-α local en utilisant le module de transfert radiatif de la raie Lyman α et montrons que ce flux modifie le spectre de puissance du signal à 21-cm, particuliérement pendant la première phase de la réionisation. Nous trouvons également que le signal est en absorption forte ce qui ne peut pas être simulé avec les hypothèses habituelles.<br /> Même une faible quantité de rayons X peut affecter l'état physique du gaz neutre dans le milieu inter galactique, puisque ceux-ci ont un libre parcours moyen très long. Le chauffage par les rayons X est un facteur important pour évaluer la puissance du signal à 21-cm dans les régions neutres, aussi avons nous inclus le ray-tracing correspondant dans LICORICE. Nous constatons que les rayons X ont besoin de temps pour élever la température du gaz au-dessus de la température du CMB. Ainsi, avec un niveau de rayons X raisonnable, nous continuons à observer le signal en absorption pendant la premi`ere phase de la réionisation.<br />De plus, nous étudions la formation des galaxies dans un contexte cosmologique. Le but est d'explorer la physique des galaxies primordiales en considérant la rétroaction radiative sur l'hydrodynamique. Nos résultats préliminaires montrent que le calcul de l'ionisation hors équilibre modifie la température du gaz diffus, et son histoire d'accrétion.