Can we trust elemental abundances derived in late-type giants with the classical 1D stellar atmosphere models?

Abstract

Abstract.We compare the abundances of various chemical species as derived with 3D hydrodynamical and classical 1D stellar atmosphere codes in a late-type giant characterized by T eff = 3640 K, log g = 1.0, [M/H]= 0.0. For this particular set of atmospheric parameters the 3D-1D abundance differences are generally small for neutral atoms and molecules but they may reach up to 0.3-0.4 dex in case of ions. The 3D-1D differences generally become increasingly more negative at higher excitation potentials and are typically largest in the optical wavelength range. Their sign can be both positive and negative, and depends on the excitation potential and wavelength of a given spectral line. While our results obtained with this particular late-type giant model suggest that 1D stellar atmosphere models may be safe to use with neutral atoms and molecules, care should be taken if they are exploited with ions.Keywords. Stars: late-type, stars: atmospheres, stars: abundances, convection Current 3-dimensional hydrodynamical codes have taken stellar atmosphere modeling to a new level of realism, making it possible to assess the influence of various nonstationary phenomena (e.g., convection) on the observable properties of various classes of stars. It has been recently demonstrated by Collet et al. (2007) that in the domain of latetype giants, which are important tracers of intermediate age and old stellar populations, the use of 1D classical and 3D hydrodynamical stellar atmosphere models may result in substantially different elemental abundances, especially at [M/H] −2.0. However, since Collet et al. (2007) have studied only stars located close to the base of RGB it is not clear whether similar conclusions would apply at lower effective temperatures and gravities.In this work we used a model of a considerably cooler late-type giant (T eff = 3640 K, log g = 1.0, [M/H]= 0.0) to assess the 3D-1D abundance differences at low effective temperatures. For this purpose we synthesized a number of artificial spectral lines at different wavelengths and with different excitation potentials, corresponding to various neutral atoms, ions and molecules. 3D and 1D stellar atmosphere models used in the spectral line synthesis were calculated with the CO 5 BOLD and LHD stellar atmosphere codes using identical atmospheric parameters (Freytag et al. 2002;Freytag et al. 2003;Wedemeyer et al. 2004). The CO 5 BOLD simulation was run on a Cartesian grid of 150x150x151 grid points (xyz, respectively). Both models shared identical equation of state and opacities. 209at https://www.cambridge.org/core/terms. https://doi

Simulations of stellar convection with CO5BOLD

“…The aforementioned investigations focused on dwarf or subgiant stars. There are also ongoing efforts to extend the application of CO5BOLD simulations to giant stars including studies of their abundances [172,173,174].…”