Stellar lithium abundances are the subject of many current debates in astrophysics (e.g., Thorburn 1996, Crane 1995, Spite & Pallavicini 1995). This paper continues the study of Li i line formation reported by Kiselman (1997), hereafter referred to as Paper I. The aim is to improve the understanding of lithium abundance determinations of solar-type stars by investigating the formation of lithium lines outside the classical realm of plane-parallel homogeneous photospheres and local thermodynamic equilibrium (LTE). It is also intended to test the idea put forward in Paper I that spatially resolved solar spectroscopy can be used to test NLTE results in less model-dependent ways than when just line profiles in integrated light are used. At the same time it tests the limits of such observational work since the lines studied are very weak.
Paper I was motivated by the proposal of Kurucz (1995) that lithium abundances derived for extremely metal-poor solar-type stars using 1D photospheric models and assuming LTE may be underestimated by as much as 1 dex due to 3D NLTE effects. The idea was that the transparency of "cold" regions in metal-poor photospheres would cause all lithium there to be ionised by ultraviolet radiation from "hot" regions. This would lead to a very weak line in the resulting mean spectrum, since the contribution to the spectral line from the hot regions is already small.
No such large effect of granulation on abundance determinations has yet been demonstrated by observations or detailed simulations. The standard behaviour of lines over the solar granulation pattern seems to be that lines - irrespective of the ionisation stage - get stronger in bright granules and weaker in darker intergranular lanes (Steffen 1989, Kiselman 1994). Thus one can expect that the influence on abundance ratios, which are derived from line ratios, will not be excessively large (Holweger et al. 1990).
The investigation of Li i line formation in a 3D model of solar granulation in Paper I demonstrated that the Kurucz mechanism at least does not work in solar granulation. This is also the conclusion of Uitenbroek (1998) who used a different NLTE treatment but the same kind of granulation model.
As regards metal-poor stars, the final verdict must of course wait until we know more about the inhomogeneity properties of their photospheres. This paper will only concern the solar Li i lines and is organised as follows. First the simulations are discussed in somewhat more detail than in Paper I. Then the solar observations are described, and finally the observations are compared with the simulation results.
© European Southern Observatory (ESO) 1998
Online publication: April 20, 1998