Stars on the asymptotic giant branch (AGB) lose a significant fraction of their mass to their surroundings through stellar winds. As a result, they are surrounded by circumstellar shells of gas and dust. This stellar mass loss replenishes and enriches the interstellar medium (ISM) with the products of stellar nucleosynthesis, progressively increasing its metallicity and thereby driving galactic chemical evolution. In this thesis I present a comprehensive study of oxygen-rich (O-rich) AGB stars and red supergiants (RSG) observed with the Spitzer Infrared Spectrograph and Infrared Space Observatory Short Wavelength Spectrometer in the Milky Way, the Large and Small Magellanic Clouds, and Galactic globular clusters; focusing on the composition of the dust in the circumstellar envelopes surrounding these stars. Combining spectroscopic and photometric observations with the GRAMS grid of radiative transfer models to derive (dust) mass-loss rates, I detect crystalline silicates in stars with dust mass-loss rates which span over a factor of 1000, down to rates of ~10^{-9} Msun/yr. Detections of crystalline silicates are more prevalent in higher mass-loss rate objects, and our results indicate that the dust mass-loss rate has a greater influence on the crystalline fraction than the gas mass-loss rate, suggesting that thermal annealing of amorphous silicate grains is the primary formation mechanism of crystalline silicates in such environments rather than the direct condensation of crystalline silicates from the gas phase. I also find that metallicity influences the composition of crystalline silicates, with enstatite seen increasingly at low metallicity, while forsterite becomes depleted at these metallicities due to the different chemical composition of the gas. To trace the evolution of alumina and silicate dust along the AGB, I present an alternative grid of MODUST radiative transfer-models for a range of dust compositions, mass-loss rates, dust shell inner radii and stellar parameters. Our analysis shows that the AKARI [11]-[15] versus [3.2]-[7] colour is a robust indicator of the fractional abundance of alumina in O-rich AGB stars. From the modelling, I show that a grain mixture consisting primarily of amorphous silicates, with contributions from amorphous alumina and metallic iron provides a good fit to the observed spectra of O-rich AGB stars in the LMC. In agreement with previous studies, we find a correlation between the dust composition and mass-loss rate; the lower the mass-loss rate the higher the percentage of alumina in the shell. Finally, I present mid-infrared observations of the Local Group dwarf elliptical galaxy M32; where I find a large population of dust-enshrouded stars. These observations will act as a pathfinder for future observations with the JWST and SPICA.

The weekly averaged near-infrared fluxes for 2652 stars were extracted from the cold and warm era all-sky maps of the Diffuse Infrared Background Experiment (DIRBE). Since the DIRBE program only archived the individual Calibrated Infrared Observations for the 10 month cold era mission, the weekly averaged fluxes were all that were available for the warm era. The steps required to extract stellar fluxes are described as are the adjustments that were necessary to correct the results for several systematic effects. The observations are at a cadence of once a week for 3.6 years (∼1300 days), providing continuous sampling on variable stars that span the entire period for the longest fundamental pulsators. The stars are divided into three categories: those with large amplitude of variability, smaller amplitude variables, and sources whose near-infrared brightness do not vary according to our classification criteria. We show examples of the results and the value of the added baseline in determining the phase lag between the visible and infrared.

To search for phase lags in the optical-infrared light curves of asymptotic giant branch stars, we have compared infrared data from the COBE DIRBE satellite with optical light curves from the AAVSO and other sources. We found 17 examples of phase lags between the times of infrared and optical maximum, and 4 stars with no observed lags. There is a clear difference between the Mira variables and the semiregular variables in the sample, with the maximum in the optical preceding that in the near-infrared in the Miras, while in most of the semiregular variables no lags are observed. Comparison to published theoretical models indicates that the phase lags in the Miras are due to strong titanium oxide absorption in the visual at stellar maximum, and suggests that Miras pulsate in the fundamental mode, while at least some semiregular variables are first-overtone pulsators. There is a clear optical-near-infrared phase lag in the carbon-rich Mira V CrB; this is likely due to C 2 and CN absorption variations in the optical.

We present the COBE DIRBE Point Source Catalog, an all-sky catalog containing infrared photometry in 10 infrared bands from 1.25 to 240 μm for 11,788 of the brightest near and mid-infrared point sources in the sky. Since DIRBE had excellent temporal coverage (100-1900 independent measurements per object during the 10 month cryogenic mission), the Catalog also contains information about variability at each wavelength, including amplitudes of variation observed during the mission. Since the DIRBE spatial resolution is relatively poor (0°.7), we have carefully investigated the question of confusion and have flagged sources with infrared-bright companions within the DIRBE beam. In addition, we filtered the DIRBE light curves, for data points affected by companions outside of the main DIRBE beam but within the "sky" portion of the scan. At high Galactic latitudes (|b| > 5°), the Catalog contains essentially all of the unconfused sources with flux densities greater than 90, 60, 60, 50, 90, and 165 Jy at 1.25, 2.2, 3.5, 4.9, 12, and 25 μm, respectively, corresponding to magnitude limits of approximately 3.1, 2.6, 1.7, 1.3, -1.3, and -3.5. At longer wavelengths and in the Galactic plane, the completeness is less certain because of the large DIRBE beam and possible contributions from extended emission. The Catalog also contains the names of the sources in other catalogs, their spectral types, variability types, and whether or not the sources are known OH/IR stars. We discuss a few remarkable objects in the Catalog, including the extremely red object OH 231.8+4.2 (QX Pup), an asymptotic giant branch star in transition to a protoplanetary nebula, which has a DIRBE 25 μm amplitude of 0.29 ± 0.07 mag.

The notion that low- to intermediate-mass young stellar objects (YSOs) gain mass at a constant rate during the early stages of their evolution appears to be challenged by observations of YSOs suffering sudden increases of the rate at which they gain mass from their circumstellar discs. Also, this idea that stars spend most of their lifetime with a low accretion rate and gain most of their final mass during short-lived episodes of high accretion bursts, helps to solve some long-standing problems in stellar evolution. The original classification of eruptive variables divides them in two separate subclasses known as FU Orionis stars (FUors) and EX Lupi stars (EXors). In this classical view FUors are at an early evolutionary stage and are still gaining mass from their parent envelopes, whilst EXors are thought to be older objects only surrounded by an accretion disc. The problem with this classical view is that it excludes younger protostars which have higher accretion rates but are too deeply embedded in circumstellar matter to be observed at optical wavelengths. Optically invisible protostars have been observed to display large variability in the near-infrared. These and some recent discoveries of new eruptive variables, show characteristics that can be attributed to both of the optically-defined subclasses of eruptive variables. The new objects have been proposed to be part of a new class of eruptive variables. However, a more accepted scenario is that in fact the original classes only represent two extremes of the same phenomena. In this sense eruptive variability could be explained as arising from one physical mechanism, i.e. unsteady accretion, where a variation in the parameters of such mechanism can cause the different characteristics observed in the members of this class. With the aim of studying the incidence of episodic accretion among young stellar objects, and to characterize the nature of these eruptive variables we searched for high amplitude variability in two multi-epoch infrared surveys: the UKIDSS Galactic Plane Survey (GPS) and the Vista Variables in the Via Lactea (VVV). In order to further investigate the nature of the selected variable stars, we use photometric information arising from public surveys at near- to farinfrared wavelengths. In addition we have performed spectroscopic and photometric follow-up for a large subset of the samples arising from GPS and VVV. We analyse the widely separated two-epoch K-band photometry in the 5th, 7th and 8th data releases of the UKIDSS Galactic Plane Survey. We find 71 stars with _K > 1 mag, including 2 previously known OH/IR stars and a Nova. Even though the mid-plane is mostly excluded from the dataset, we find the majority (66%) of our sample to be within known star forming regions (SFRs), with two large concentrations in the Serpens OB2 association (11 stars) and the Cygnus-X complex (27 stars). The analysis of the multi-epoch K-band photometry of 2010-2012 data from VVV covering the Galactic disc at |b| < 1◦ yields 816 high amplitude variables, which include known variables of different classes such as high mass X-ray binaries, Novae and eclipsing binaries among others. Remarkably, 65% of the sample are found concentrated towards areas of star formation, similar to the results from GPS. In both surveys, sources in SFRs show spectral energy distributions (SEDs) that support classification as YSOs. This indicates that YSOs dominate the Galactic population of high amplitude infrared variable stars at low luminosities and therefore likely dominate the total high amplitude population. Spectroscopic follow-up allows us to confirm the pre-main sequence nature of several GPS and VVV Objects. Most objects in both samples show spectroscopic signatures that can be attributed to YSOs undergoing high states of accretion, such as veiling of photospheric features and CO emission, or show FUor-like spectra. We also find a large fraction of objects with 2.12 μm H2 emission that can be explained as arising from shock-excited emission caused by molecular outflows. Whether these molecular outflows are related to outbursts events cannot be confirmed from our data. Adding the GPS and VVV spectroscopic results, we find that between 6 and 14 objects are new additions to the FUor class from their close resemblance to the near-infrared spectra of FUors, and at least 23 more objects are new additions to the eruptive variable class. For most of these we are unable to classify them into any of the original definitions for this variable class. In any case, we are adding up to 37 new stars to the eruptive variable class which would double the current number of known objects. We note that most objects are found to be deeply embedded optically invisible stars, thus increasing the number of objects belonging to this subclass by a much larger factor. In general, objects in our samples which are found to be likely eruptive variable stars show a mixture of characteristics that can be attributed to both of the optically-defined classes. This agrees well with the recent discoveries in the literature. Finally, we are able to derive a first rough estimate on the incidence of episodic accretion among class I YSOs in the star-forming complex G305. We find that _ 9% of such objects are in a state of high accretion. This number is in agreement with previous theoretical and observational estimates among class I YSOs.

The influence of macroclumps for free–free spectral energy distributions (SEDs) of ionized winds is considered. The goal is to emphasize distinctions between microclumping and macroclumping effects. Microclumping can alter SED slopes and flux levels if the volume filling factor of the clumps varies with radius; however, the modifications are independent of the clump geometry. To what extent does macroclumping alter SED slopes and flux levels? In addressing the question, two specific types of macroclump geometries are explored: shell fragments (pancake-shaped) and spherical clumps. Analytic and semi-analytic results are derived in the limiting case that clumps never obscure one another. Numerical calculations based on a porosity formalism is used when clumps do overlap. Under the assumptions of a constant expansion, isothermal, and fixed ionization wind, the fragment model leads to results that are essentially identical to the microclumping result. Mass-loss rate determinations are not affected by porosity effects for shell fragments. By contrast, spherical clumps can lead to a reduction in long-wavelength fluxes, but the reductions are only significant for extreme volume filling factors.

We present an analysis of observations using the Spitzer Space Telescope and the James Clerk Maxwell Telescope of deeply embedded protostars in the Perseus Giant Molecular Cloud. Building on the results of Jørgensen et al. (2007), we attempt to characterize the physical properties of these deeply embedded protostars, discovered due to their extremely red near infrared colours and their proximity to protostellar cores detected at 850 μm. Using a grid of radiative transfer models by Robitaille et al. (2006), we fit the observed fluxes of each source, and build statistical descriptions of the best fits. We also use simple one dimensional analytic approximations to the protostars in order to determine the physical size and mass of the protostellar envelope, and use these 1D models to provide a goodness-of-fit criterion when considering the model grid fits to the Perseus sources. We find that it is possible to create red [3.6]-[4.5] and [8.0]-[24] colours by inflating the inner envelope radius, as well as by observing embedded protostars through the bipolar outflows. The majority of the deeply embedded protostars, however, are well fit by models seen at intermediate inclinations, with outflow cavity opening angles < 30o, and scattering of photons off of the cavity walls produces the red colours. We also discuss other results of the SED fitting.

Large spectroscopic surveys are trailblazing endeavours in the study of stellar archaeology and near eld cosmology. Access to homogeneous databases of thousands of stellar spectra allow for a detailed and statistically satisfying look into the chemical abundance distribution of our Galaxy and its surrounding satellites, ultimately working towards a better understanding of galactic chemical evolution. This thesis presents the work of three new studies at the current frontier of stellar archaeology. Through the rst look at carbon-enhanced metal-poor (CEMP) stars using H-band spectra, six new CEMP stars and another seven likely candidates were found within the APOGEE database following Data Release 12. These stars have chemical compositions typical of metal-poor halo stars, however the alpha-abundances of two stars indicate possible origins in an accreted dwarf galaxy. A lack of heavy element spectral lines impedes further sub-classi cation of these CEMP stars, however, based on radial velocity scatter, we predict most are not CEMP-s stars which are typically found in binary systems. This preliminary investigation warrants optical observations to con rm the stellar parameters and low metallicities of these stars, to determine the heavy-element abundance ratios and improve the precision in the derived abundances, and to examine their CEMP sub-classi cations. Additionally, the rst results for the spectroscopic follow up to the Pristine survey are presented. Using a sample of 149 stars, a success rate of 70% for finding stars with [Fe/H]<-2.5 and 22% for finding stars with [Fe/H]<-3.0 is reported, significantly higher than other surveys that typically report success rates of 3-4% for recovering stars with [Fe/H]<-3.0. Finally, the new spectral analysis tool StarNet is introduced. A deep neural network architecture is used to examine both synthetic stellar spectra and SDSS-III APOGEE spectral data and can produce the stellar parameters of temperature, gravity, and metallicity with similar or better precision as the APOGEE pipeline values when trained directly with the APOGEE spectra. StarNet is capable of being trained on synthetic data as well, and is able to reproduce the stellar parameters for both synthetic and APOGEE spectra, including low signal-to-noise spectra, with similar precision to training on the APOGEE spectra itself. The residuals between StarNet predictions and APOGEE DR13 parameters are similar to or better than the di erences between the APOGEE DR13 results and optical high resolution spectral analyses for a subset of benchmark stars. While developed using the APOGEE spectral database (real spectra and corresponding ASSET synthetic data with similar normalization functions), StarNet should be applicable to other large spectroscopic surveys like Pristine. / Graduate

Conselho Nacional de Desenvolvimento Científico e Tecnológico / We present a multi-wavelength study of the stellar content of 11 optical/infrared Southern Galactic Hii regions with 10 h < α(J2000) < −17 h and −65° < δ(J2000) < −35°. Nine optical objects with no published or uncertain distances were examined in order to identify their ionising sources and to determine their distances, whereas young stellar objects (YSOs) and main sequence ionising stars were photometrically classified in the two infrared clusters. We carried out optical spectroscopy and UBV photometry of the stellar content of these Hii regions and obtained the distance of individual stars by spectroscopic parallax. To avoid using a fixed value for the total-to-selective extinction ratio RV , the reddening AV was determined directly by the colour-difference method. We classified as types O or B 24 out of the 31 stars for which optical spectra were obtained. In particular, we identified two new O stars in RCW98 and RCW99. The values for reddening obtained correspond to a mean hRV i = 3.44, which is about 10% higher of the mean value found for field stars over all directions of the Galaxy. For the Hii regions NGC3503, NGC6334, RCW55, RCW87, RCW98 e RCW99, we obtained more precise distances calculated as the median of the spectroscopic parallaxes obtained for two to six different stars in each nebulae, with an internal dispersion of less than 5%. Among the objects more obscured in the Norma region, we analysed the stellar content of the East part of the large complex of Hii regions GAL331.5-00.1, which contains eight bright extended radio sources. This study comprises the infrared clusters [DBS2003] 156 and [DBS2003] 157, respectively associated to Hii regions GAL331.11-00.51 and GAL331.31-00.34. In order to isolate the brightest 2MASS sources of their unresolved companions, we carried out JHK photometry with resolution better than 2MASS data. 47 Near-infrared (NIR) sources with Ks-band excess were identified following usual methods. Other 70 YSOs were also identified by using Mid-IR on-line data of the GLIMPSE survey. The search for radial-velocity measurements in the literature and the similarity between the stellar population explored have indicated the two regions as physically associated. With the determination of the spectroscopic parallaxes of four O-type and two B-type stars spread over the both clusters, this hypothesis was verified. The parallaxes of these 6 stars returned very compatible distances (hdhelioci = 3.30 ± 0.29 kpc). The Near- and Mid-IR counterparts of the IRAS source 16085-5138 was found close to the field of [DBS2003] 157. This source has showed typical colours of a Ultra-compact ii region (UCHii) and spectral index α = 3.6 between 2 and 25 μm, which is typical of YSOs immersed in protostellar envelopes. A lower limit to the bolometric luminosity of the protostar embedded was computed as L = 7.7×103L⊙ (M = 10M⊙), which corresponds to a BO-B1 zero-age star. The cluster [DBS2003] 157 was found to be spread over all ∼4′ × 4′ region demarcated by a intense shell-like dust emission, where a secondary massive star formation is going on, likely as a result of the interaction between this dust and stellar winds. / Apresentamos aqui um estudo multibanda do conteúdo estelar de onze regiões Hii do Hemisfério sul com 10 h < α(J2000) < −17 h e −65◦ < δ(J2000) < −35◦. Em 9 objetos ópticos, com valores de distância discrepantes ou sem nenhuma determinação anterior publicada, visamos identificar as fontes ionizantes e determinar suas distâncias, enquanto que, no estudo de dois outros objetos no infravermelho, buscamos não apenas identificar as fontes ionizantes, mas também objetos estelares jovens (YSOs) em seus campos. Nos primeiros, realizamos espectroscopia no óptico e fotometria UBV de seus conteúdos estelares e determinamos as distâncias das estrelas massivas por meio da paralaxe espectroscópica. A fim de evitar o uso de um valor fixo para a razão entre a extinção total e a seletiva à banda V , RV , a extinção AV na direção de cada estrela foi determinada diretamente através no método das diferenças de cor. De um total de 31 estrelas espectroscopicamente estudas, 24 foram classificadas como tipos O ou B, sendo duas novas estrelas tipo O encontradas em RCW98 e RCW99. As estimativas de AV implicaram um valor médio de hRV i = 3.44. Este resultado supera em 10% o valor médio encontrado para estrelas de campo sobre todas as direções da Galáxia. Para as regiões Hii NGC3503, NGC6334, RCW55, RCW87, RCW98 e RCW99, distâncias mais precisas foram estimadas como o valor mediano da paralaxe espectroscópica de 2 a 6 diferentes estrelas ionizantes, resultando em uma dispersão interna menor que 5%. Dentre os objetos mais obscurecidos pela extinção interestelar da região de Norma, analisamos o conte´udo estelar da região leste do grande complexo de regiões Hii brilhantes em radiofrequências GAL331.5-00.1. A área estudada engloba os aglomerados infravermelhos [DBS2003] 156 e [DBS2003] 157, respectivamente associados às regiões Hii GAL331.11-00.51 e GAL331.31-00.34. Observações fotométricas nas bandas J, H e Ks, com mais alta resolução que a fotometria 2MASS, foram realizadas nas direções desses objetos, o que possibilitou isolar as fontes infravermelhas mais brilhantes de estrelas companheiras não resolvidas e selecionar potenciais estrelas ionizantes para subsequente espectroscopia. 47 fontes com excesso de emissão intrínseca na banda Ks, típico em YSOs, foram identificadas seguindo os métodos usuais de análise da fotometria no IR próximo (NIR). Outros 70 YSOs foram identificados no IR médio (Mid-IR) usando dados do survey GLIMPSE. A pesquisa por medidas de velocidade radial na direção desses dois objetos e a semelhança das populações estelares indicou que as duas sub-regiões estudas deveriam estar fisicamente associadas. Esta hipótese foi reafirmada com a determinação da paralaxe espectroscópica de 4 estrelas tipo O e outras 2 tipo B nos dois aglomerados, que retornaram valores de distâncias heliocêntricas bastante compatíveis (hdhelioci = 3.30 ± 0.29 kpc). A contrapartida nos NIR e Mid-IR da fonte IRAS 16085-5138 foi encontrada junto ao aglomerado [DBS2003] 157. Esta fonte apresentou cores típicas de uma região Hii Ultracompacta (UCHii) e índice espectral entre 2 e 25 μm de α = 3.6, típico de YSO imerso em um envelope protoestelar. Um limite inferior para a luminosidade bolométrica da protoestrela embebida foi estimado em L = 7.7×103L⊙ (M = 10M⊙), o que corresponde a uma estrela de idade zero na faixa de BO-B1. O aglomerado [DBS2003] 157 mostrou-se estar espalhado sobre toda uma região de ∼4′ ×4′, demarcada por intensa emissão de poeira quente e espacialmente distribuída como uma nuvem em forma de concha.