[Paper Review] 3D Hydrodynamical Simulations of Surface Convection in Red Giant Stars. Impact on spectral line formation and abundance analysis

R. Collet, M. Asplund+1 more|ArXiv.org|Mar 26, 2007

Stellar, planetary, and galactic studies134 citations

TL;DR

This study uses 3D hydrodynamical simulations of red giant star atmospheres to investigate how realistic convective motions affect spectral line formation and elemental abundance determinations. It finds that 3D models predict significantly cooler upper atmospheric layers than 1D models, leading to stronger neutral species and molecular lines under LTE, resulting in 3D–1D abundance corrections of −0.5 to −1.0 dex for C, N, O, and Fe in metal-poor giants at [Fe/H] ≈ −3.

ABSTRACT

We investigate the impact of 3D hydrodynamical model atmospheres of red giant stars at different metallicities on the formation of spectral lines of a number of ions and molecules. We carry out realistic 3D simulations of surface convection in red giant stars with varying stellar parameters. We use the simulations as time-dependent hydrodynamical model stellar atmospheres to compute atomic (Li, O, Na, Mg, Ca, Fe) and molecular (CH, NH, OH) spectral lines under the assumption of local thermodynamic equilibrium (LTE). We compare the line strengths computed in 3D with the results of analogous line formation calculations for 1D, hydrostatic, plane-parallel MARCS model atmospheres in order to estimate the impact of 3D models on the derivation of elemental abundances. The temperature and density inhomogeneities and correlated velocities in 3D models, as well as the differences between the 1D and mean 3D structures significantly affect the predicted line strengths. Under the assumption of LTE, the low atmospheric temperatures of very metal-poor 3D model atmospheres cause the lines from neutral species and molecules to appear stronger than in 1D. Therefore, elemental abundances derived from these lines using 3D models are significantly lower than according to 1D analyses. Differences between 3D and 1D abundances of C, N, and O derived from CH, NH, and OH weak low-excitation lines are found to be in the range -0.5 dex to -1.0 dex for the the red giant stars at [Fe/H]=-3 considered here. At this metallicity, large negative corrections (about -0.8 dex) are also found for weak low-excitation Fe I lines. We caution, however, that departures from LTE might be significant for these and other elements and comparable to the effects due to stellar granulation.

Motivation & Objective

• To assess the impact of 3D hydrodynamical model atmospheres on spectral line formation in red giant stars.
• To quantify the differences in elemental abundance determinations between 3D and classical 1D model atmospheres.
• To investigate how temperature inhomogeneities, velocity fields, and stratification differences in 3D models affect line strengths under LTE.
• To evaluate the significance of 3D effects on abundance analysis in metal-poor red giants, especially for C, N, O, and Fe.
• To identify systematic uncertainties in 3D abundance analysis, particularly regarding non-LTE effects and scattering approximations.

Proposed method

• Performing ab initio 3D hydrodynamical simulations of surface convection in red giant stars across varying effective temperatures and metallicities.
• Using the resulting time-dependent 3D model atmospheres as input for spectral line formation calculations under local thermodynamic equilibrium (LTE).
• Computing spectral line strengths for atomic (Li i, O i, Na i, Mg i, Ca i, Fe i, Fe ii) and molecular (CH, NH, OH) lines in both 3D and 1D model atmospheres.
• Conducting a differential comparison of line strengths between 3D and 1D models to derive 3D–1D abundance corrections.
• Applying the MARCS 1D model atmosphere code for comparison with 3D results, assuming identical stellar parameters.
• Assessing the impact of scattering approximated as absorption and potential non-LTE effects on the derived corrections.

Experimental results

Research questions

• RQ1 How do 3D hydrodynamical model atmospheres alter the predicted strengths of spectral lines compared to 1D models in red giant stars?
• RQ2 What is the magnitude and direction of 3D–1D corrections to elemental abundances derived from weak low-excitation lines of C, N, O, and Fe in metal-poor red giants?
• RQ3 To what extent do temperature inhomogeneities and correlated velocity fields in 3D models influence line formation and abundance determinations?
• RQ4 How do departures from LTE and scattering approximations affect the reliability of 3D abundance corrections?
• RQ5 Can 3D models resolve systematic errors in 1D abundance analyses related to hydrostatic equilibrium and mixing-length theory?

Key findings

• 3D model atmospheres exhibit significantly cooler upper layers than 1D models due to temperature inhomogeneities and correlated velocity fields.
• Spectral lines of neutral species and molecules (CH, NH, OH) appear stronger in 3D models than in 1D models under LTE due to the cooler temperatures in the upper atmospheric layers.
• For red giants at [Fe/H] ≈ −3, 3D–1D abundance corrections for C, N, and O derived from weak low-excitation CH, NH, and OH lines range from −0.5 to −1.0 dex.
• Large negative corrections of approximately −0.8 dex are found for Fe i lines at the same metallicity, indicating that 1D analyses overestimate Fe abundances.
• The assumption of LTE may introduce significant errors, as non-LTE effects—particularly over-ionization in Fe—can counteract the granulation-induced line strengthening, suggesting that non-LTE effects are comparable in magnitude to 3D effects.
• Systematic uncertainties in 3D abundance analysis arise from approximating scattering as absorption and from neglecting non-LTE effects, both of which can affect temperature structure and flux predictions in the UV.