The importance of collision broadening of weak lines in stellar spectra

Several formalisms for computing the collision broadening of weak iron lines are compared. These include approximations to Van der Waals broadening, empirical enhancement factors, the WIDTH6 approximation, and the Anstee & O’Mara calculations. Abundances are computed for a set of red Fe I lines using solar data, to illustrate the effects of selecting one or other damping formalism. The main points of this research are: 1. even weak lines (log (W/λ) = −5) lacking strong wings are sensitive to the choice of damping, the abundances inferred from themvarying by asmuch as 0.1 in dwarfs (but by less in giants) depending on the adopted formalism and excitation potential. In slightly stronger lines, having log (W/λ) = −4.8, the errors may reach 0.2 dex; 2. damping errors can depend strongly on both excitation potential and equivalent width, and could mislead an analyst into adopting an inappropriate effective temperature (in error by∼100 K or more) and microturbulence. In particular, the WIDTH6 formalism is probably unsuitable for many optical lines having χ > 2 eV. 3. the effects of an incorrect damping choice can be reduced, though not eliminated, by adjusting themicroturbulence, but only for elements with many lines spanning a large range of equivalent widths. In practice, this restricts the adjustment to neutral iron lines, leaving the analysis of other species open to potentially large errors; 4. until reliable damping values are available for many lines of many species, the benefits of having high S/N spectra and gf values accurate to a few percent will seldom be realized in abundances accurate to a few ×0.01 dex; 5. the poor understanding of microturbulence and the need to quantify it empirically during an abundance analysis represent significant deficiencies in our understanding of stellar absorption line physics, and limit the accuracy of stellar abundances. Send offprint requests to: S. Ryan ([email protected])