The primordial nature of the C-19 stellar stream

Abstract

Stellar streams, remnants of compact star systems stretched out by the tidal forces of the Milky Way, offer a unique way to study stellar populations that formed billions of years ago. A particularly unique stream is C-19, the most metal-poor stellar stream known at less than a thousandth of the Sun's metallicity. The nature of C-19 is not yet clear, with properties that resemble both star clusters and ultra faint dwarf galaxies, yet in either case its extremely low metallicity indicates very early star formation, <1 Gyr after the Big Bang. Here, we present the first detailed study on the nature of C-19 based on the chemical abundances of 14 member stars from high-resolution spectroscopy. These reveal that C-19 formed stars in an early, rapid, and prolific star formation event, with mild inhomogeneous mixing of elements produced in massive stars. There is otherwise no evidence for subsequent star formation, multiple stellar populations, nor chemical evolution. Although C-19 is currently disrupted in the Milky Way halo, it offers a rare and complementary window into the details of star formation and chemical evolution in the early universe, ideal for comparisons with current studies of primordial star formation in the high-redshift universe.

Figures (6)

• Figure 1: The upper panel shows fourteen C-19 members from Table \ref{['tab:targets']} in the celestial coordinates ($\alpha$, $\delta$), where the transparent pink band represents the disk region with $|b|< 10^{\circ}$ in Galactic coordinates. The eight members in the outer stream are colored in the same way as Fig. \ref{['fig:srba']}, one of which is separated by the disk in the North. The lower left panel shows the velocities of the C-19 members along $\delta$. The particles from the simulation (see Metho are plotted as gray dots, and those close to the observed C-19 stream is highlighted in purple, following the stream orbit, shown as solid black line. The lower right panel shows the color-magnitude diagram of the C-19 members as well as the candidates without radial velocity information. The solid black line here represents a 12-Gyr isochrone with [Fe/H]$=-2.2$ at the distance of 18 kpc, from the standard PARSEC model bressan2012.
• Figure 2: Light element NLTE abundances for C-19 compared to metal-poor ([Fe/H]$<-2.5$) stars in the Milky Way and five globular clusters: two metal-poor GCs (M15/white, M92/darkgrey), and three low mass GCs (Ter 8/lavender, IC4499/lightblue, NGC6535/peach, and each with purple edges). Three stars in the core of C-19 (C19KLM) are from published analysis of Gemini/GRACES spectra (grey diamonds). References are in Section \ref{['secA1']}. Our measurements for the standard star HD122563 are shown as a large filled grey circle with black edges and errorbar.
• Figure 3: Sr and Ba NLTE abundances plotted as a function of [Ba/Fe] (top panel) and [Ba/H] (bottom panel, which includes a line of fixed [Sr/H] $=-3.8$ in light blue). C-19 stellar abundances are compared to metal-poor stars in the Milky Way (SAGA Suda2017 in light grey, and HLi2022 in medium grey), and the two metal-poor GCs M15 Sobeck11Garcia2024 and M92 Kirby2023. Abundances from individual stars in UFDs are also shown, collected from the SAGA database (Suda2017, references in Methods) and the GHOULS survey (Dovgal2025, medium blue). Three UFDs with $>3$ stars with Sr and Ba each are highlighted: Bootes I (black edges from SAGA), Com Ber (purple edges from Sitnova2021), and Ret II (bright blue from SAGA) which is offset due to a rare r-process event Ji2023Ret2. We note that the system Tuc III (brown) resembles the metal-poor globular clusters more closely than UFDs. References are in Section \ref{['secA1']}. Our measurements for the standard star HD122563 are shown as a large filled grey circle with errorbar.
• Figure 4: The mean chemistry of C-19 Venn2025 (blue stars) compared to Pop III core-collapse supernova yields Heger2012 (grey) and two Pop II rapidly rotating massive star models Limongi2018 (orange). All models are within 3$\sigma$ of the best fitting model (with $\chi^2$=0.47). Residuals for each element between the observed and predicted abundance for each model are shown in the top plot. Right side shows the range in these best fit models in Mass and Explosion Energy, and where the two rapidly rotating massive star models (v$_{\rm rot}=300$ km/s) are shown with orange boxes. The general conclusion is that normal CCSN of stars near 20 M$_\odot$ are sufficient to explain the available chemical abundances in C19.
• Figure 5: Bottom-left panel: Two-dimensional likelihood function of the mean and dispersion of the [Fe/H] metallicity for C-19 based on the sample of 8 new stars presented here (magneta contours), on the 6 literature stars listed in Table \ref{['tab:params']} (blue contours), and for the combined data sets that are independent (thick black contours). A dot represents the favoured model in each case. Top and right-hand panels: One-dimensional marginalized likelihood function for the mean and dispersion, respectively. From the combined sample of 14 C-19 stars, we infer that C-19 has a mean metallicity $\langle[\textrm{Fe}/\textrm{H}]\rangle=-3.31^{+0.07}_{-0.08}$ and a metallicity dispersion $\sigma_{\mathrm{[Fe/H]}}<0.18$ dex at the 95% confidence level.