The metal-poor tail of the APOGEE survey I. Uncovering [Fe/H] < -2.5 stars from the inner Galaxy to the Magellanic Clouds

TL;DR

This study addresses the challenge of identifying metal-poor stars in APOGEE data beyond the pipeline's metallicity floor of $[M/H]=-2.5$, extending the search into dust-obscured regions like the inner Galaxy and the Magellanic Clouds. The authors combine APOGEE uncalibrated parameters, calibrated Teff and log g, high $S/N$, and careful spectral inspection, then validate candidates against multiple external literature metallicity catalogs. They derive distances with BaSTI isochrones and compute long-term orbits with AGAMA in a barred Milky Way potential, identifying 289 high-confidence VMP red giant stars (including 16 Magellanic Cloud members, 14 inner-Galaxy stars, and 13 inner-Galaxy halo interlopers) and flagging 34 dwarfs and 16 Magellanic members for follow-up. The results reveal predominantly halo-like dynamics with substantial orbital chaos in the inner Galaxy, demonstrating the feasibility of using H-band surveys to discover metal-poor populations across diverse Galactic environments and informing future surveys; Paper II will provide detailed Mg and Si abundances for these targets, and data are released via CDS.

Abstract

In the search for metal-poor stars, large spectroscopic surveys are an invaluable tool. However, the spectra of metal-poor stars can be difficult to analyse because of the relative lack of available lines, which can also lead to misclassification. We aim to identify the stars observed by the APOGEE survey that are below the metallicity limit of APOGEE's analysis. For the highest confidence candidates, we classify the orbital properties of the stars to investigate whether their orbital distribution matches what we would expect for stars that are this metal poor and to select especially interesting targets for spectroscopic follow-up purposes. We examined the properties derived by APOGEE for metal-poor stars from the literature to find signatures of stars with a metallicity below the range of the grid used for spectral analysis. The calibrated APOGEE stellar parameters provide a clear signature of the most metal-poor stars in the survey, indicated by null values for their metallicities while having effective temperatures and surface gravities determined by the pipeline. From comparison with the literature, we find that, within a temperature range of 3700 - 6700 K and above a threshold of S/N > 30, the vast majority of APOGEE stars without calibrated metallicities are very metal poor. The radial velocities provided by APOGEE, Gaia DR3 positions and astrometry along with spectrophotometric distances derived in this work allowed for the computation of their orbits. In this work, we select 289 very metal-poor red giant stars (likely below = -2.5) from the APOGEE results. Our sample contains 16 very metal-poor member candidates of the Magellanic Clouds, 14 very metal-poor stars with orbits confined to the inner Galaxy, and 13 inner Galaxy halo interlopers.

Figures (16)

• Figure 1: Metallicity distribution functions for SAGA stars with literature $\rm [Fe/H]<-2.5$. Top panel: uncalibrated metallicities from APOGEE. Bottom panel: calibrated APOGEE metallicities. The number of null values for each metallicity is shown with the same y-axis.
• Figure 2: Kiel diagram of all APOGEE stars, colour-coded by their calibrated $\textsc{M\_H}$ values, plotted with the lowest metallicity stars on top. The stars with null metallicity measurements are plotted in greyscale: stars with $\textsc{TEFF}$ too high or too low to reliably get metallicity measurements are shown in black; stars with S/N < 30 are shown in dark grey; stars with S/N > 30 are shown in light grey.
• Figure 3: Normalised histograms of the S/N for all APOGEE stars (shown in red) and for the stars with calibrated $\textsc{TEFF}$ but without calibrated metallicity (shown in grey). For very low S/N values the stars without metallicity are clearly overrepresented compared to the overall distribution. At S/N $\gtrsim$ 30 the distribution flattens, suggesting that the reason for failing to derive a calibrated metallicity is not low S/N.
• Figure 4: Spectral view of APOGEE spectra for the 326 candidate VMP stars in this work (grey lines), after cleaning the sample. Vertical thin lines across the wavelength space connect to artefacts in some of these spectra that are not affecting the overall analysis. Also shown are the averaged APOGEE spectra for red giants with temperatures and gravities in a range comparable to that of the bulk of our sample (4100 K $< \rm T_{eff} <$ 5200 K and 0.5 $< \log(g) <$ 3.5) and a distinct range in metallicities. The comparison clearly shows that our selected sample stars are indeed very metal poor.
• Figure 5: Normalised histogram of the S/N on a log scale of the candidate stars, shown in grey, and of the subset of candidates with values from the literature sources in Sect. \ref{['sec:comp']}, including the additional sources in section \ref{['sec:more_lit']}, split into photometric and spectroscopic metallicities. The overall S/N distribution of APOGEE is shown as a red outline, with the right side y-axis tracking their numbers.