In the course of the ESO "Large Programme" First Stars, accurate measurements of the abundances of the first stars in our Galaxy have been obtained for giant stars, providing encouraging results. For example, the evolution of the abundances of iron and chromium was measured simultaneously. The excellent correlation displayed shows the small amplitude of random errors in these measurements. However the slope of Cr/Fe with Fe/H, displayed in Figure 1, (first discovered 14 years ago by McWilliam and collaborators), is not easily explained by nucleosynthesis, which suggests that the determinations are affected by some kind of systematic error.
Subsequently the team performed the same measurements in a sample of dwarf stars, and the results are somewhat different, notably for Cr.
Namely, if one relies on the measurement of neutral Cr lines, the slope is a litle smaller for the dwarf stars and the Cr/Fe ratio somewhat larger. If one, instead, relies on ionised Cr (only one line measurable in giant stars) the slope essentially disappears and the ratio Cr/Fe is equal to the solar value, suggesting that the Cr and Fe evolve in lockstep. This result suggests that the formation of Cr lines is not adequately computed when converting measurements to abundances or abundance ratios. All the computations have been performed making the hypothesis of local thermodynamic equilibrium (LTE). Such hypothesis is often sufficient, but not always, and obviously not in the present case. To avoid using this hypothesis it would be necessary to know exactly the energy levels and transition probabilities of the chrome atom : let us hope that atomic physicists will provide such data in the near future. The GEPI astronomers are also working to avoid a further approximation in the abundance calculation, by taking into account all the motions which occur in the stellar atmosphere. A large grid of three dimensional hydrodynamic simulations of stellar atmospheres has been computed and employed in the analysis of dwarfs stars. Extension of the grid to giant stars, which are computationally more demanding is ongoing. This has been made possible by a Marie Curie Excellence grant, which supports the CIFIST Team at GEPI. Atomic physics, the study of departures from LTE in stellar atmospheres and the hydrodynamical simulations of stellar atmospheres will contribute to a better understanding of nucleosynthesis.