Abstract
Atomic diffusion must be taken into account in the computations of stellar structure and evolution as it is a direct consequence of the fact that stars are self-gravitating spheres composed of a mixture of different gases (the chemical elements). The stellar equilibrium leads to internal gradients of pressure, density and temperature as well as an upward radiative transfer which produces a selective effect on the elements (in most cases dominated by the competition between gravitational settling and radiative acceleration).The interactions between atomic diffusion and well-known hydrodynamical processes like dynamical convection and mass loss have been studied for a long time. An important process was however forgotten in these computations. This is the double-diffusive (or fingering or thermohaline) convection induced by unstable μ-gradients, which can be produced by the local accumulation of heavy elements inside stars due to radiative acceleration. Contrary to the other hydrodynamical processes, fingering convection is not arbitrarily added in the computations. It is directly induced by atomic diffusion itself and cannot be avoided. It is thus very important to add this hydrodynamical process in stellar evolution modelling, which has never been done before our work. A similar effect occurs below the convective zone in case of accretion of heavy matter onto a star.We studied the accretion-induced fingering convection in the case of the stellar system 16 Cygni. We studied the properties of these two stars by computing models with the Toulouse Geneva Evolution Code (TGEC). We included the Brown et al. 2013 prescription for the computations of fingering convection in the code. We computed oscillation frequencies of these models using the PULSE code to compare it with Kepler observations. We found that if 2/3 of Earth mass is accreted at the beginning of the main sequence (on 16 Cyg B model), the accretion-induced fingering convection mixes the star deep enough to destroy the lithium and obtain the observed difference between 16 Cyg A and B.We studied the heavy element accumulation and the induced fingering convection in the case of Am stars. In these stars, peculiar surface abundances are observed (compared to the sun). This peculiarity is related to the effect of atomic diffusion, very important in these types of stars. However, atomic diffusion alone leads to abundance variations which are too large and one way to reproduce the observed abundance quantitatively is to assume mixing deep enough inside the star. We computed models including atomic diffusion (with radiative acceleration) and fingering convection with this prescription using the TGEC code. We find than this process may change the internal structure of the stars, and also the surface abundances. We also included fingering convection and the accretion process in the Montreal/Montpellier code. We modified some parts of this code (e.g. turbulence profiles) to compare the results obtained with the two codes. We computed some models and I found that the results are quite similar.We determined the stellar parameters of the star 94 Ceti (by using similar seismic computations as for 16 Cyg A and B) using ground-based observations. This star has a mass of 1.44 MΘ and is a good target to study the effect of radiative accelerations (which occur for masses larger than 1.2-1.3 MΘ). We also compared models with full atmosphere with the observations to determine the impact on oscillation frequencies.We worked on metal poor halo stars for which a dispersion of lithium surface abundance is observed for very small metallicities. We studied the possibility of an accretion of matter that can trigger fingering convection and destroy lithium.