Estrelas anãs brancas são o último estágio evolutivo observável de mais de 95% das estrelas e também um resultado comum na evolução de estrelas binárias. O estudo de anãs brancas é, portanto, uma ferramenta poderosa na compreensão da evolução de estrelas simples e binárias, da função de massa inicial local, e da perda de massa após a sequência principal, levando-nos a uma melhor compreensão do histórico de formação e evolução estelar de diferentes populações. Para que esses estudos sejam possíveis, é necessária uma amostra grande e preferencialmente completa de anãs brancas, cobrindo todo o intervalo de parâmetros físicos. A maneira mais simples de obter isso é utilizando dados de grandes projetos de mapeamento. O Sloan Digital Sky Survey já permitiu o aumento do número de anãs brancas conhecidas em cinco vezes até o data release 10. Neste trabalho, estendemos a busca por anãs brancas aos novos objetos no data release 12, descobrindo 3 157 novas anãs brancas e 1 349 novas subanãs. Pela primeira vez, estendemos essa busca para log g < 6.5, correspondente a M < 0.3 M⊙. Anãs brancas abaixo desse limite de massa não podem formar-se em um tempo de Hubble; contudo, se a estrela é parte de um sistema de binárias próximas, a perda de massa pode ser tão intensa que a anã branca resultante tem massa inferior ao limite por evolução simples. Esses objetos são chamados anãs brancas de massa extremamente-baixa (ELMs, do inglês extremely-low mass white dwarfs) Elas têm Teff < 20 000 K e 5.0 . log g . 6.5 e espectros muito similares a estrelas A de sequência principal. Menos de cem são conhecidas, e a maioria dos objetos foi descoberta tendo em conta um critério de seleção tendencioso, que excluiu ELMs frias (Teff < 9 000 K) e com mais baixa massa (M . 0.15 M⊙), tornando difícil verificar modelos e compreender as propriedades das ELMs como classe. Nós identificamos milhares de objetos cujas propriedades físicas, temperatura efetiva e log g, estão no intervalo de produtos de evolução binária, como as ELMs. Nós os chamamos de sdAs, referindo-nos ao seu log g que as coloca abaixo da sequência principal e seu espectro dominado por hidrogênio. As sdAs parecem conter populações estelares sobrepostas, e encontramos que 7% são mais compatíveis com (pre-)ELMs do que com objetos de sequência principal. Obtivemos espectroscopia resolvida temporalmente para 26 objetos e pudemos confirmar que 15 estão em binárias. Um objeto faz parte de um sistema eclipsante, enquanto outro é pulsante — o oitavo membro da classe de ELMs pulsantes. Outros seis objetos também apresentaram pulsações em nossas observações, cinco desses estão na vizinhança da faixa de instabilidade das ELMs. Com estes resultados, aumentamos a população de ELMs por um fator de 20%, elevando a fração de ELMs de 4 para 20%, um valor que é consistente com as previsões de modelos evolucionários. / White dwarf stars are the final observable evolutionary state of over 95% of stars and also a common outcome in binary evolution. Therefore, studying white dwarfs is a powerful tool to understand both single and binary stellar evolution, local initial mass function, and post-main sequence mass loss, leading us to a better comprehension of the history of stellar formation and evolution of different stellar populations. In order to make this type of studies possible, a large and preferably complete sample of white dwarf stars, covering the whole range of physical parameters, is required. The simplest way to achieve that is to take advantage of data provided by large surveys. The Sloan Digital Sky Survey has already allowed the increase of the number of known white dwarf stars fivefold up to its data release 10. In this work, we extended the search for white dwarfs to the new objects in the data release 12, discovering 3 157 new white dwarfs and 1 349 new subdwarfs. For the first time, we have extended this search to log g < 6.5, corresponding to M < 0.3 M⊙. White dwarfs below this mass limit cannot be formed through single evolution within a Hubble time; however, if the star is part of a close binary system, the mass loss of the system may be so intense that the resulting white dwarf has mass below the single evolution limit. These objects are known as extremely-low mass white dwarfs (ELMs) They show Teff < 20 000 K and 5.0 . log g . 6.5 and spectra very similar to main sequence A stars. Less than a hundred of them are known, and most objects were discovered relying on biased selection criteria, that excluded cool (Teff < 9 000 K), lowermass (M . 0.15 M⊙) ELMs, making it difficult to validate the models and comprehend the properties of the ELMs as a class. We have identified thousands of objects whose physical properties, effective temperature and surface gravity, place them in the range of by-products of binary interaction such as the ELMs. We have called them sdAs, referring to their sub-main sequence log g and hydrogen dominated spectra. They seem to be composed of overlapping stellar populations, and we found that at least 7% are more likely ELMs or their precursors, the pre-ELMs, than main sequence stars. Obtaining time-resolved spectroscopy for 26 objects, we could confirm 15 to be in close binaries. One of them is also an eclipsing system, while another is a pulsator — the eighth member of the pulsating ELM class. Other six new pulsators were found as part of our follow-up, five of them in the vicinity of the ELM instability strip. With these results, we increase the population of ELMs by 20%, raising the fraction of cool ELMs from 4 to 20%, which is consistent with the predictions from the evolutionary models.

In this thesis I present the search for ultracool dwarf companions to main sequence stars, subgiants and white dwarfs. The ultracool dwarfs identified here are benchmark objects, with known ages and distances. The online data archives, the two micron all sky survey (2MASS) and SuperCOSMOS were searched for ultracool companions to white dwarfs, where one M9 1 companion to a DA white dwarf is spectroscopically confirmed as the widest separated system of its kind known to date. The age of the M9 1 is constrained to a minium age of 1.94Gyrs, based on the estimated age of the white dwarf from a spectroscopically derived Teff and log g and an initial-final mass relation. This search was extended using the next generation surveys, the sloan digital sky survey (SDSS) and the UK infrared deep sky survey (UKIDSS), where potential white dwarf + ultracool dwarf binary systems from this search are presented. A handful of these candidate systems were followed-up with second epoch near infrared (NIR) imaging. A new white dwarf with a spectroscopic M4 companion and a possible wide tertiary ultracool component is here confirmed. Also undertaken was a pilot imaging survey in the NIR, to search for ultracool companions to subgiants in the southern hemisphere using the Anglo-Australian telescope. The candidates from that search, as well as the subsequent follow-up of systems through second epoch NIR/optical imaging and methane imaging are presented. No systems are confirmed from the current data but a number of good candidates remain to be followed-up and look encouraging. A search for widely separated ultracool objects selected from 2MASS as companions to Hipparcos main-sequence stars was also undertaken. 16 candidate systems were revealed, five of which had been previously identified and two new L0 2 companions are here confirmed, as companions to the F5V spectroscopic system HD120005 and the M dwarf GD 605. The properties of HD120005C were calculated using the DUSTY and COND models from the Lyon group, and the age of the systems were inferred from the primary members. For GD 605B no age constraint could be placed due to the lack of information available about the primary, but HD120005C has an estimated age of 2-4Gyr. In the final part of this thesis I investigate correlations with NIR broadband colours (J - H, H - K and J - K) with respect to properties, Teff , log g and [Fe/H] for the benchmark ultracool dwarfs, both confirmed from the searches undertaken in this work and those available from the literature. This resulted in an observed correlation with NIR colour and Teff, which is presented here. I find no correlation however with NIR colours and log g or [Fe/H], due in part to a lack of suitable benchmarks. I show that despite the current lack of good benchmark objects, this work has the potential to allow UCD properties to be measured from observable characteristics, and suggest that expanding this study should reveal many more benchmarks where true correlation between properties and observables can be better investigated.

HD 11112 is an old, Sun-like star that has a long-term radial velocity (RV) trend indicative of a massive companion on a wide orbit. Here we present direct images of the source responsible for the trend using the Magellan Adaptive Optics system. We detect the object (HD 11112B) at a separation of 2 2 (100 au) at multiple wavelengths spanning 0.6-4 mu m. and show that it is most likely a gravitationally bound cool white dwarf. Modeling its spectral energy distribution suggests that its mass is 0.9-1.1M(circle dot), which corresponds to very high eccentricity, near edge-on orbits from a. Markov chain Monte Carlo analysis of the RV and imaging data together. The total age of the white dwarf is > 2 sigma, which is discrepant with that of the primary star under most assumptions. The problem can be resolved if the white dwarf progenitor was initially a double white dwarf binary that then merged into the observed high-mass white dwarf. HD 11112B is a unique and intriguing benchmark object that can be used to calibrate atmospheric and evolutionary models of cool white dwarfs and should thus continue to be monitored by RV and direct imaging over the coming years.

I present the development of two near infrared proper motion pipelines for high resolution near infrared data from UKIDSS and the VISTA VVV survey. The UKIDSS pipeline is capable of accuracies of order 5-10 mas yr⁻¹ for bright sources with the largest epoch baselines (~ 8 years). The VVV pipeline reaches 1-2 mas yr⁻¹ proper motion precision at the bright end and parallax measurements at ~ 1 mas precision. It will be possible to improve upon the VVV astrometric precision due to increases in data volume and further pipeline development. I have used the proper motion pipelines to generate three near infrared proper motion catalogues of the UKIDSS LAS and GPS and the VVV survey. The LAS proper motion catalogue covers 1500 deg2 at high Galactic latitudes and contains approximately 15 million sources with two J band epochs. The GPS proper motion catalogue covers 1500 deg2 of the northern Galactic plane and contains approximately 400 million sources with two K band epochs. The VVV proper motion catalogue covers 560 deg2 of the Galactic bulge and disc and contains approximately 200 million sources with between 50 and 150 Ks band epochs. I have also produced a preliminary 5σ parallax catalogue containing 3403 VVV sources. The LAS and GPS proper motion catalogues have been used by myself and other authors to identify and study many new examples of high proper motion stars, brown dwarfs, ul-tracool dwarf benchmark candidates, cool white dwarfs, substellar subdwarfs and nearby sources within < 25 pc. These catalogues remain far from fully exploited and will be a useful resource for future research by the astronomical community. Exploitation of the VVV proper motion catalogue is still in its infancy, yet it has already generated large numbers of new high proper motion sources. These include new brown dwarf candi-dates, important benchmark objects, and nearby sources which have previously avoided detection. Parallax results from the VVV pipeline will be useful to improve low mass star/ultracool dwarf luminosity functions, significantly increasing the numbers of brown dwarfs with known parallaxes and illustrates how general purpose multi-epoch wide area surveys can generate parallaxes. Finally, I discuss the long term usefulness of such catalogues in the Gaia era and how they might be exploited in the future.

Ce mémoire présente, pour la première fois, des analyses détaillées d’étoiles naines blanches montrant à la fois des éléments lourds et de forts champs magnétiques. Ces analyses sont effectuées à l’aide d’un tout nouveau code d’atmosphère pouvant calculer la position et la force des raies de tous les éléments (du carbone au cuivre) en régime Paschen-Back pour une géométrie de notre choix. Dans un premier temps, nous décrivons l'effet d'un champ magnétique sur les niveaux d'énergie atomiques, par l'effet Zeeman en champ faible puis en régime Pashen-Back lorsque l'approche perturbative n'est plus valide. Nous explorons ensuite l’effet de la géométrie du champ magnétique, notamment les différences entre les modèles calculés en supposant un champ magnétique uniforme (ou constant dans l'espace) et ceux ayant une géométrie dipolaire. Certaines études ont récemment affirmé qu’en présence d’un champ magnétique intense le mouvement convectif était fortement atténué, de sorte que l’atmosphère devenait pratiquement radiative (Tremblay et al., 2015). Nous explorons brièvement l’impact que peut avoir la suppression du transport d’énergie convective sur les résultats d’analyses d’étoiles magnétiques. Dans le but d'améliorer les analyses d'étoiles magnétiques, nous avons implémentés certains changements aux modèles d'atmosphères utilisés, tel que l'ajout du traitement magnétique des raies métalliques. Nos nouveaux outils nous permermetterons d'analyser de manière rigoureuse, pour la première fois, les étoiles G165-7, J2105+0900 et LHS 2534, trois étoiles magnétiques avec présence d'éléments lourds. Nous étudions leurs paramètres atmosphériques tels que la température effective, gravité de surface et abondances de métaux à partir de modèles standards ainsi qu’à partir de modèles radiatifs ou la convection a été artificiellement inhibée par la présence d’un champ magnétique intense. Nous observons que nos modèles riches en hydrogène sans convection ne reproduisent plus les observations, contrairement à ceux riches en hélium où la convection se produit hors de la région de formation des raies (soit plus profondément). Nous observons finalement qu'un champ magnétique dipolaire centré avec l'étoile ne reproduit pas bien les profils des raies d'un même multiplet, puisque les composantes sigma (celles à gauche et à droite de la raie centrale) prédites sont trop larges. Les observations spectroscopiques sont mieux reproduites à partir de modèles ayant un champ magnétique de surface constant, ou avec un dipôle toujours fortement décalé selon l’axe -z, indiquant que la géométrie réelle du champ magnétique est probablement plus complexe qu’un simple dipôle. / We present, for the first time, detailed analyses of white dwarf stars showing heavy metal lines and large magnetic fields. These analyses are carried out with a new atmosphere code able to compute the position and strength of lines for all elements (from carbon to copper) in the Paschen-Back regime for an arbitrary geometry. Firstly, we describe the effect of a magnetic field on the atomic energy levels, first with the Zeeman effect for weak fields then in the Paschen-Back regime when the pertubative approach is no longer valid. We then explore the effect of the magnetic field geometry, especially the differences between models calculated assuming a constant field and those with a dipolar geometry. Some studies recently suggested that in the presence of an intense magnetic field, the convective movement is strongly inhibited, renderingthe atmosphere mostly radiative (Tremblay et al., 2015). We briefly explore the impact the suppression of the convective energy transport can have on the analyses' results of magnetic stars. In order to carry out a thorough analysis of some magnetic stars, we have applied changes to the atmospheric models used, for example by adding magnetic treatment of metallic lines. With these new tools, we analyse for the first time, in a rigorous manner, the stars G165-7, J2105+0900 and LHS 2534, three magnetic stars showing heavy elements. We study their atmospheric parameters such as the effective temperature, surface gravity and metal abundances, and then explore the possibility of having pure radiative atmospheres, where the convection would be inhibited by intense magnetic fields. We find that hydrogen rich models cannot reproduce observations, unlike those rich in helium where the convective movement takes place out of the line forming region. We finally note that a centered dipolar magnetic field is not able to reproduce the line profiles of a given multiplet, since the sigma components (those to the left and the right of the central line) predicted are too broad. Spectroscopic observations are better reproduced with constant magnetic field models, or with a strongly offset dipole in the -z axis, indicating that the actual field geometry probably is more complex than a simple dipole.

Cette thèse présente les avancées dans l'étude des étoiles naines blanches magnétiques. L'étude des étoiles naines blanches en général, le state final d'évolution de près de 95% des étoiles de notre galaxie, permet d'obtenir de l'information cruciale sur leurs progéniteurs. Se pencher sur les propriétés des champs magnétiques des étoiles naines blanches par opposition à celles de leur progéniteur, beaucoup plus difficiles à observer en partie à cause de la faible intensité du champ, est bien plus réalisable. Plusieurs hypothèses ont été formulées afin de décrire l'origine et l'évolution de ces objets atypiques. Que les champs soient d'origine fossile suite à l'évolution d'une étoile isolée et conservation du flux magnétique, ou créés lors de la rotation rapide de la fusion de deux étoiles, il nous faut bien comprendre leur résultat, les naines blanches magnétiques, avant de pouvoir retracer leur évolution. Toutefois, les modèles de naines blanches magnétiques d'aujourd'hui reposent sur un nombre d'approximation, qui ne sont valides jusqu'à une relativement faible intensité de champ. Ceci a été la motivation derrière ce projet de doctorat. Nous avons implémenté un traitement des raies d'absorption utilisant le calcul des énergies de transition de l'atome d'hydrogène et d'hélium sous champs magnétiques arbitraires. Ces calculs, fournis par le groupe de l'université de Tübingen en Allemagne, ont utilisé une approche numérique à la résolution de l'Hamiltonien de l'atome d'hydrogène et d'hélium en présence d'un champ magnétique d'intensité arbitraire. Cet ajout aux modèles d'atmosphères permet de modéliser la position et l'intensité des raies d'absorption dans le spectre des étoiles. Une procédure de discrétisation de la surface visible de l'étoile nous permet de modéliser des champs magnétiques avec une géométrie arbitraire, tel qu'un dipôle décentré et incliné par rapport à l'observateur. Ces nouveaux outils ont été utilisés afin d'étudier en premier lieu un ensemble des naines blanches magnétiques avec une atmosphère riche en hydrogène. De ces résultats, nous avons déterminé une distribution de masse de ces étoiles, avec une masse moyenne significativement plus élevée que leurs homologues non-magnétiques, tels qu'attendu par les théories d'évolution binaires. Les informations extraites des différentes caractéristiques des distributions obtenues suite à cette analyse ne nous permettent cependant pas de favoriser une hypothèse d'évolution plutôt qu'une autre. Suite à ceci, celles avec une atmosphère riche en hélium ont été les prochaines à être analysées. Nous avons ainsi déterminé rigoureusement, pour la première fois, les paramètres magnétiques de ces étoiles riches en hélium. / This thesis presents advances in the study of magnetic white dwarf stars. The study of white dwarf stars in general, the final state of evolution of nearly 95% of the stars in our galaxy, provides crucial information about their progenitors. It is much more feasible to study the magnetic field properties of white dwarf stars as opposed to their progenitor stars, which are much more difficult to observe in part because of the weak field strength. Several hypotheses have been formulated to describe the origin and evolution of these atypical objects. Whether the fields are of fossil origin following the evolution of an isolated star and conservation of magnetic flux, or created during the rapid rotation of the merger of two stars, we need to understand their result, the magnetic white dwarfs, before we can trace their evolution. However, today's models of magnetic white dwarfs rely on a number of approximations, which are only valid up to relatively low field strengths. This was the motivation behind this doctoral thesis. We have implemented a treatment of the absorption lines using the calculation of the transition energies of the hydrogen and helium atom under arbitrary magnetic fields. These calculations, provided by the group of the University of Tubingen in Germany, used a numerical approach to the resolution of the Hamiltonian of the hydrogen and helium atom in the presence of a magnetic field of arbitrary intensity. This addition to atmospheric models allows to model the position and intensity of absorption lines in the spectrum of stars. A discretization procedure of the visible surface of the star allows us to model magnetic fields with an arbitrary geometry, such as an offset and inclined dipole. These new tools have been used to study first a set of magnetic white dwarfs with a hydrogen-rich atmosphere. From these results, we have determined a mass distribution of these stars, with an average mass significantly higher than their non-magnetic counterparts, as expected by the binary evolution theories. The information extracted from the different characteristics of the distributions obtained from this analysis does not however allow us to favor one evolutionary hypothesis over another. Following this, those with a helium rich atmosphere were the next to be analyzed. We have thus determined rigorously, for the first time, the magnetic parameters of these helium-rich stars.

Le nombre de systèmes planétaires connus ne cesse de croître depuis la découverte de la première exoplanète il y a déjà quelques décennies. Cependant, ces systèmes connaîtront éventuellement la mort de leur étoile, s’ils ne l’ont pas déjà connue. Ces étoiles deviendront des naines blanches. Pourtant, peu de planètes ont été découvertes autour de naines blanches. Une théorie populaire propose la survie des planètes géantes autour de celles-ci. Il est toutefois difficile de rechercher des planètes autour de naines blanches, entre autres, à cause de leurs petits rayons et de leurs raies spectrales peu nombreuses. L’imagerie de noyaux de phase est une méthode prometteuse pour la recherche d’exoplanètes autour de naines blanches puisqu’elle permet l’exploration à des séparations angulaires très petites allant jusqu’à la moitié de la limite de diffraction. Il s’agit de détecter un signal astrophysique par l’entremise de l’annulation des erreurs de phase instrumentales produites par le télescope. Ce projet vise à évaluer les performances de l’instrument MIRI du JWST pour détecter des exoplanètes autour de naines blanches avec cette technique. Pour ce faire, les observations de quatre naines blanches ont été analysées. Ce projet a permis de découvrir deux exoplanètes candidates, soient WD 1202−232 b et WD 2105−820 b sur un échantillon de quatre naines blanches. Elles ont été détectées par imagerie directe standard, mais aussi par imagerie de noyaux de phase. Les candidates se trouvent respectivement à une séparation de 1230 ± 20 mas et 2210 ± 20 mas de leur étoile et leurs rapports de flux sont de 63 ± 2 pour le système WD 1202−232 et de 29 ± 3 pour le système WD 2105−820. Si les candidates sont confirmées, la masse de WD 1202−232 b se situerait entre 1 et 7 \(M_J\), alors que la masse de WD 2105−820 b se situerait entre 1 et 2 \(M_J\). Des analyses et des simulations ont aussi été faites dans le but de tester les performances de l’imagerie par noyaux de phase avec les observations du JWST/MIRI à différents régimes de signal sur bruit. / The number of known planetary systems has been continuously increasing since the discovery of the first exoplanet several decades ago. However, these systems will eventually experience the death of their star, if they have not already. These stars will become white dwarfs. Yet, few planets have been discovered around white dwarfs. A favored theory suggests the survival of giant planets around them. However, it is challenging to search for planets around white dwarfs, among other reasons, due to their small radii and limited number of spectral lines. Kernel phase imaging is a promising method for the search for exoplanets around white dwarfs as it allows exploration at very small angular separations down to half the diffraction limit. This involves detecting an astrophysical signal through the cancellation of instrumental phase errors produced by the telescope. This project aims to evaluate the performance of JWST’s MIRI instrument in detecting exoplanets around white dwarfs using this technique. To achieve this, the observations of four white dwarfs were analyzed. This project has led to the discovery of two candidate exoplanets, namely WD 1202−232 b and WD 2105−820 b, in a sample of four white dwarfs. They were detected through standard direct imaging but also through kernel phase imaging. The candidates are located at a separation of 1230 ± 20 mas and 2210 ± 20 mas from their star, with flux ratios of 63 ± 2 for the WD 1202−232 system and 29 ± 3 for the WD 2105−820 system. If confirmed, the mass of WD 1202−232 b would be between 1 and 7 \(M_J\), while the mass of WD 2105−820 b would be between 1 and 2 \(M_J\). Analyses and simulations were also conducted to test the performance of kernel phase imaging with JWST/MIRI observations under various signal-to-noise regimes.

Le but de cette thèse est de raffiner et de mieux comprendre l'utilisation de la méthode spectroscopique, qui compare des spectres visibles de naines blanches à atmosphère riche en hydrogène (DA) à des spectres synthétiques pour en déterminer les paramètres atmosphériques (température effective et gravité de surface). Notre approche repose principalement sur le développement de modèles de spectres améliorés, qui proviennent eux-mêmes de modèles d'atmosphère de naines blanches de type DA. Nous présentons une nouvelle grille de spectres synthétiques de DA avec la première implémentation cohérente de la théorie du gaz non-idéal de Hummer & Mihalas et de la théorie unifiée de l'élargissement Stark de Vidal, Cooper & Smith. Cela permet un traitement adéquat du chevauchement des raies de la série de Balmer, sans la nécessité d'un paramètre libre. Nous montrons que ces spectres améliorés prédisent des gravités de surface qui sont plus stables en fonction de la température effective. Nous étudions ensuite le problème de longue date des gravités élevées pour les DA froides. L'hypothèse de Bergeron et al., selon laquelle les atmosphères sont contaminées par de l'hélium, est confrontée aux observations. À l'aide de spectres haute résolution récoltés au télescope Keck à Hawaii, nous trouvons des limites supérieures sur la quantité d'hélium dans les atmosphères de près de 10 fois moindres que celles requises par le scénario de Bergeron et al. La grille de spectres conçue dans ces travaux est ensuite appliquée à une nouvelle analyse spectroscopique de l'échantillon de DA du SDSS. Notre approche minutieuse permet de définir un échantillon plus propre et d'identifier un nombre important de naines blanches binaires. Nous déterminons qu'une coupure à un rapport signal-sur-bruit S/N > 15 optimise la grandeur et la qualité de l'échantillon pour calculer la masse moyenne, pour laquelle nous trouvons une valeur de 0.613 masse solaire. Finalement, huit nouveaux modèles 3D de naines blanches utilisant un traitement d'hydrodynamique radiative de la convection sont présentés. Nous avons également calculé des modèles avec la même physique, mais avec une traitement standard 1D de la convection avec la théorie de la longueur de mélange. Un analyse différentielle entre ces deux séries de modèles montre que les modèles 3D prédisent des gravités considérablement plus basses. Nous concluons que le problème des gravités élevées dans les naines blanches DA froides est fort probablement causé par une faiblesse dans la théorie de la longueur de mélange. / The goal of this thesis is to refine and to understand better the spectroscopic method, which compares optical spectra of hydrogen-atmosphere white dwarfs (DA) with synthetic spectra to determine the atmospheric parameters (effective temperature and surface gravity). Our approach rests mainly on the development of improved model spectra, which come themselves from DA model atmospheres. We present a new grid of DA synthetic spectra with the first consistent implementation of the non-ideal gas theory of Hummer & Mihalas and the unified theory of Stark broadening from Vidal, Cooper & Smith. This allows for an adequate treatment of the quenching effects in Balmer lines, without the need of a free parameter. We show that these improved spectra predict surface gravities that are much more stable as a function of the effective temperature. We then study the long-standing problem that surface gravities in cool DA stars are significantly higher than those found in hotter DA white dwarfs. The hypothesis of Bergeron et al., according to which the atmospheres are contaminated by small amounts of helium, is constrained with observations. Using high-resolution spectra collected at the Keck in Hawaii, we find superior limits on the helium abundances in the atmospheres that are nearly 10 times lower than those required to sustain the Bergeron et al. scenario. The grid of spectra calculated in this work is then applied to a new spectroscopic analysis of the DA in the SDSS. Our careful analysis allows us to define a cleaner sample and to identify a large number of double degenerates. We find that a cutoff at a signal-to-noise ratio S/N > 15 optimizes the size and quality of the sample for computing the average mass, for which we find a value of 0.613 solar mass. Finally, eight new 3D white dwarf models with a radiative-hydrodynamics treatment of the convection are presented. We also calculated models with the same physics, except for a treatment of the convection with the standard mixing-length theory. A differential analysis between these two sets of models shows that the 3D models predict considerably lower surface gravities. We conclude that the high-log g problem in cool DA white dwarfs is caused by a weakness in the mixing-length theory.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.