Li and Be depletion in metal-poor subgiants

Stellar atmospheres constitute excellent environments to study the chemical evolution of our Galaxy. The chemical composition of these atmospheres reflects the composition of the gas from where these stars were born. As the Galaxy evolves, the composition of the gas changes from being primordial (Big-Bang nucleosynthesis) to being enriched in heavy elements (stellar and interstellar nucleosynthesis). The abundances of fragile chemical elements can be affected by stellar mixing processes. Precise lithium, beryllium and oxygen abundance determinations in old stars are presented in this thesis. These determinations are based on the analysis of the observed spectra of a sample of thirteen metal-poor subgiant stars. According to stellar mixing theories, these stars are in a stellar evolutionary stage in which mixing by convection is expected. Abundances of fragile elements like lithium and beryllium are thus expected to be affected by such mixing processes. As a consequence of this, the abundances of these elements are discussed in a dilution context. Lithium and beryllium abundances are compared with the abundances of stars with similar characteristics but in a less evolved stellar phase so that mixing processes have not acted yet. As expected, our abundances seem to be depleted following reasonably well the standard predictions. Stellar abundances of oxygen should give an estimate of the oxygen contribution of core-collapse supernovae to the interstellar medium. However, there is poor agreement among the abundances determined from different atomic or molecular indicators in general. Abundances coming from three different indicators are compared in this thesis. The abundances determined from the O I infrared triplet lines at 777.1-5 nm give the poorest agreement among the three indicators. The abundances based on OH ultraviolet lines around 310 nm are lower for the subgiants in comparison with previous studies of main-sequence stars, becoming even lower than values based on the O I forbidden line at 630.03 nm. Still the most reliable indicator appears to be the O I forbidden line which suggests a plateau-like or only slowly increasing [O/Fe] towards lower [Fe/H]. In addition, the line formation of the Be II ultraviolet resonance lines at 313.0-1 nm, commonly used for abundance determinations purposes, is investigated under non-local thermodynamic equilibrium. We find that the common assumption of local thermodynamic equilibrium typically gives systematic errors of about 0.1 dex.