Gaia FGK Benchmark Stars: Selecting Infrared Lines for Abundance Determination

Technical Summary: Gaia FGK Benchmark Stars: Selecting Infrared Lines for Abundance Determination

Problem Statement
While optical abundance determinations benefit from extensively validated line lists and modeling frameworks, infrared (IR) measurements face significantly larger uncertainties. These limitations stem from incomplete atomic and molecular data, the quality of available laboratory data (particularly oscillator strengths and collisional broadening parameters), and the challenges of mitigating telluric contamination in ground-based observations. Existing IR line lists, such as those used by the APOGEE survey, are often astrophysically calibrated to match specific survey observations rather than anchored in laboratory data, limiting their reproducibility across different instruments and stellar populations. Furthermore, the identification of reliable absorption features in the near-infrared (NIR) Y, J, and H bands is complicated by severe line blending, molecular contamination in cooler stars, and the scarcity of benchmark datasets compared to the optical regime.

Methodology
The authors present a homogeneous, quantitative, and reproducible framework for selecting atomic absorption lines in the Y (9800–10800 Å), J (11800–13200 Å), and H (15000–17500 Å) bands. The study utilizes high-resolution (R ∼ 90,000) CRIRES+ spectra from the European Southern Observatory for six Gaia FGK Benchmark Stars (GBS), covering a wide range of effective temperatures ($T_{\text{eff}}$), surface gravities ($\log g$), and metallicities ([Fe/H]), from F dwarfs to M giants.

The selection process relies exclusively on laboratory atomic data from the Vienna Atomic Line Database (VALD3) and synthetic spectra generated using MARCS LTE models. The authors employ a five-step decision-tree filtering system to evaluate each transition independently across all six stars:

1. Depth: Lines must have a central depth ($d$) $\ge 0.03$ in both synthetic and observed spectra to ensure sensitivity above the noise floor (typically S/N > 38).
2. Saturation: Lines are rejected if they enter the saturation regime, defined by a local curve-of-growth slope $< 0.5$ and a core flux change $< 0.5\%$ for a 0.2 dex abundance variation.
3. Purity: The degree of blending is quantified by the purity parameter $P = \text{EW}_1 / \text{EW}_2$, where $\text{EW}_1$ and $\text{EW}_2$ are equivalent widths measured in narrow (30 km s$^{-1}$) and wide (60 km s$^{-1}$) velocity windows. Lines with $P < 0.5$ are rejected; those with $0.5 \le P < 0.75$ are flagged as "undecided" (U), while $P \ge 0.75$ are accepted (Y).
4. Goodness of Fit (GoF): Observed and synthetic profiles are compared using statistical metrics: reduced chi-square ($\chi^2_{\text{red}}$), Root Mean Square Error (RMSE), and Median Absolute Deviation (MAD). Thresholds are adjusted differentially for each band to account for varying telluric residuals (e.g., stricter limits for the cleaner Y band, more tolerant limits for the H band).

A line is considered "robust" only if it satisfies all criteria (final flag sequence YYYY) across the full range of stellar parameters in the sample.

Key Results
The study identifies a specific set of robust atomic transitions that remain consistent across the diverse stellar sample:

• $\alpha$-elements: Transitions of Mg I, Si I, and Ca I consistently meet all robustness criteria.
• Iron-peak elements: Several Fe I transitions are identified as robust.
• Neutron-capture elements: Among the species explored, only Sr II provides lines that consistently meet the requirements, primarily in the Y band for F and G dwarfs. Most other neutron-capture candidates were excluded due to blending or insufficient depth/purity.
• Exclusions: Many potential lines, particularly those of neutron-capture elements and certain species in cooler M giants (like $\gamma$ Sge), were rejected due to severe blending, molecular contamination, or sensitivity to model assumptions.

The authors cross-matched their robust list against the APOGEE atomic linelist and recent solar-type studies (Şentürk et al. 2024; Marfil et al. 2020). While there is significant overlap (e.g., 33 common transitions with APOGEE for G-dwarfs), the authors' list includes additional robust transitions absent from external compilations, reflecting their stricter, laboratory-based selection criteria and broader parameter coverage.

Significance and Claims
The paper claims to provide a "transparent and reproducible foundation" for near-infrared line validation. By anchoring the selection to laboratory data rather than astrophysical fine-tuning, the authors argue that their resulting line list offers a uniform basis for abundance determination that is less dependent on specific survey pipelines or instrument characteristics.

The study emphasizes that while the framework is robust, it acknowledges inherent limitations:

• Cool Giants: The methodology faces reduced predictive power for M giants (e.g., $\gamma$ Sge) where molecular absorption dominates and continuum definition is difficult.
• NLTE Effects: The analysis relies on LTE models; the authors note that IR lines may suffer from larger Non-LTE deviations, particularly for strong lines, which are not fully corrected in this work.
• Neutron-Capture Elements: The difficulty in finding clean, robust lines for neutron-capture species in the IR is highlighted, suggesting that future work may require relaxed thresholds or tailored abundance scaling strategies.

Ultimately, the paper positions itself as a step toward homogeneous chemical abundance analyses in the NIR, offering a vetted set of lines that can be used with confidence across different stellar types without the need for object-dependent fine-tuning.