Résumé
The availability of large atomic data bases has made it possible to calculate stellar evolution models taking into detailed account the abundance variations of all important contributors to opacity. In a first step, in addition to nuclear reactions, the atomic diffusion, radiative accelerations and opacity are continuously calculated during evolution taking the abundance changes of 28 species into account. This leads to the first self consistent main sequence stellar evolution models. In A and F stars ( M ≥ 1.5 M ʘ ) an iron peak convection zone is shown to appear at a temperature of 200000 K. The calculated surface abundance anomalies, that follow without any arbitrary parameter, are very similar to those observed in AmFm stars in open clusters except that they are larger by a factor of about 3. The second step, is then to introduce a competing hydrodynamical process. To reduce the calculated anomalies to the observed ones, turbulence has been introduced. It is found that the mixed zone must be about 5 times deeper than the iron convection zone. Detailed comparisons to a few AmFm stars have been carried out. The determination of the abundance anomalies of a large number of atomic species (20 to 30 are probably accessible) makes it possible to constrain stellar hydrodynamics. In clusters, the original abundances and age may be known and the accurate determination of surface abundances may constrain turbulence, mass loss and differential rotation when the required atomic data bases are available and used for the modeling of particle transport in stellar evolution.