Revisiting the near infrared Calcium triplet as metallicity indicator

TL;DR

The paper tackles refining Ca II triplet based metallicity indicators by incorporating Gaia $G$-band luminosity as an additional proxy and extending the calibration across a wide metallicity and age range. It develops a Python-based, lmfit/MCMC–driven pipeline to measure CaT line strengths via Gaussian–Lorentzian fits and defines the CaT index $\\Sigma Ca$ as the sum of the three CaT line strengths. A nonlinear calibration is derived relating $[Fe/H]$ to $\\Sigma Ca$ and four luminosity indicators $M_V, M_I, M_{K_s}, M_G$ with coefficients estimated by lmfit and emcee, yielding residuals around $\\sim 0.2$ dex over $-4 \\\lesssim [Fe/H] \\\lesssim +0.15$ and ages $\\gtrsim 200$ Myr. The study finds good agreement with previous work at the metal-poor end but substantial differences at higher metallicities due to updated line-strength measurements and reference abundances, notably for NGC 6791, and provides a publicly available code for CaT analysis to support Gaia-era metallicity determinations.

Abstract

The near-infrared Calcium II Triplet (CaT), around 850nm, is a key metallicity indicator for red giant stars. We present a revised [Fe/H] calibration as a function of CaT line strengths and four luminosity indicators, including the $Gaia$ $G$-band, together with the classical $V$, $I$, and $K_s$ bandpasses. For this purpose, we used a sample of 366 red giant stars belonging to 25 globular and open clusters, complemented by 52 extremely metal-poor field giant stars. The CaT line strengths are determined by fitting Gaussian-Lorentzian combination profiles using the Python lmfit package, which utilises the algorithms implemented therein. The derived calibration is valid for a wide metallicity range, $-4$\,dex$ \lesssim \mathrm{[Fe/H]} \lesssim +0.15$, and for ages older than $\sim$200 Myr. In addition, we performed a detailed assessment of how factors such as spectral resolution, spectral quality (expressed through the signal-to-noise ratio), and the algorithms used to constrain the line profiles affect the measured line strengths and the resulting metallicities.

Figures (6)

• Figure 1: Comparison between the C13's strengths derived with the new Python implementation but using the Levenberg–Marquardt algorithm, for each line. Points are colour-coded as a function of the star metallicity, as noted in the right sidebar. Point shapes denote different sources, as labelled in the legend. The dashed lines of equal strength are added for reference in black.
• Figure 2: Contribution of the strengths of 849.8 nm (top-left), 854.2 nm (top-right), and 866.2 nm (bottom-left) to the global CaT index, $\Sigma$ Ca. Bottom-right panel show the behaviour of the $W_{854.2}$/$W_{866.2}$ ratio. Median (dashed lines) and standard deviation (shadow regions) shown in the bottom-right corner of each panel have been computed applying a three-sigma clipping (open circles show rejected points).
• Figure 3: Comparison of the strengths determined for each CaT line (first three left columns) and the global $\Sigma Ca$ index (right) columns determined from synthetic spectra (see text for details) of different resolutions (top row), signal-to-noise ratios (S/N, middle row) and, and using different fitting algorithms (bottom row). As a reference, we use the values determined from synthetic spectra with a spectral resolution of 8,500 using the same methodology, Nelder-Mead plus emcee package, used in the observed spectra.
• Figure 4: The relationship of the CaT index ($\Sigma Ca$) with the luminosity indicators $M_V$, $M_I$, $M_{K_s}$ and $M_G$, respectively. The colour bar represents the metallicity range for each star. Different point shapes represent three categories of different tracers, as indicated in the legend.
• Figure 5: Differences between the reference metallicities and the values derived from the new calibrations obtained here for the four luminosity indicators used (bottom panels), colour coded as a function of metallicity. The top panels show the global distributions of these residuals (black) and in different metallicity ranges (different colours). The statistic of these distributions are summarised in Table \ref{['tab:histogram_res']}.