Atomic Diffusion in Old Stars : Testing parameter degeneracies

The predicted primordial lithium abundance differs from observations of unevolved halo stars on the Spite plateau by a factor two to three. Surface depletion due to atomic diffusion has been suggested as a cause of this so-called cosmological lithium problem. Evolutionary abundance trends indicative of atomic diffusion have previously been identified in the metal-poor globular cluster NGC 6397 ([Fe/H] = -2), with stellar parameters deduced spectroscopically in a self-consistent manner. Abundances of five elements (Li, Mg, Ca, Ti, and Fe) were found to be in agreement with stellar structure models including the effects of atomic diffusion and a free-parameter description of turbulent mixing at the lowest efficiency compatible with the flatness of the Spite plateau. It is our aim to evaluate the interplay of modelling assumptions and theoretical predictions under various priors, e.g. the independent age determination using the white dwarf cooling sequence, and the high efficiency of turbulent mixing recently found compatible with halo field stars. We perform self-consistent spectroscopic abundance analyses at an expanded effective temperature scale inspired by results of new photometric calibrations from the infrared flux method. The resulting abundances are compared to predictions in a grid of theoretical isochrones, chosen in light of the priors for age and efficiency of turbulent mixing. We find that the observed abundance trends are not artefacts of the effective temperature scale, as it cannot be arbitrarily modified to flatten all trends. The inferred abundance trends seem to be in agreement with predictions for an age compatible with the white dwarf cooling sequence, and a limited range of weak turbulent mixing. The inferred initial lithium abundance of these stars is merely 30 % lower than the primordial abundance, discrepant at 1.5 standard deviations. Hence, a stellar solution to the cosmological lithium problem is still within reach