A Bimodal Metallicity Distribution Function in the Ultra-Faint Dwarf Galaxy Reticulum II

TL;DR

This paper presents a deep spectroscopic MDF for Reticulum II by measuring metallicities for 167 member stars, including MSTO stars, using Ca II K line strengths and a Beers 1999 calibration with careful zero-point alignment. A likelihood analysis shows Ret II’s MDF is bimodal, with peaks near $[ ext{Fe/H}] \approx -3.02$ and $[ ext{Fe/H}] \approx -2.08$, separated by about 0.94 dex, suggesting two distinct star formation episodes with a ~3 Gyr gap. Mixing-model fits favor a Gaussian+Gaussian interpretation, supporting a two-burst SFH and indicating that star formation continued after reionization in this ultra-faint system. The results tie the metallicity structure to an age-metallicity relation and the broader context of r-process enrichment, implying environment and internal feedback shaped Reticulum II’s chemical evolution and offering a benchmark for interpreting MDFs in the smallest galaxies.

Abstract

Star formation in ultra-faint dwarf galaxies (UFDs, $M_* <10^5M_\odot$) is suppressed by reionization, but may not be completely quenched. The metallicity distribution function (MDF) of stars in ultra-faint dwarf galaxies could show these signatures of reionization. However, past studies of UFD MDFs have been limited, because there are only a few dozen red giant branch (RGB) stars in such low-mass galaxies. We present low-resolution Magellan/IMACS spectroscopy of 167 stars in the UFD Reticulum II ($M_* \approx 3000 M_\odot$), increasing the number of stellar metallicities by 6.5 times and resulting in the most populated spectroscopic metallicity distribution function of any UFD. This is possible because we determined the first spectroscopic metallicities of main sequence turn-off stars in any UFD. The MDF of Reticulum II is clearly a bimodal distribution, displaying two peaks with about $80\%$ of the stars in the metal-poor peak at $\rm[Fe/H]=-3.0$ and $20\%$ of the stars in the more metal-rich peak at $\rm[Fe/H]=-2.1$. Such a large metallicity gap can be explained by Type Ia supernova enrichment during a long quiescent period. This supports the currently-favored two-burst star formation history for Reticulum II and shows that such low-mass galaxies clearly can form stars after reionization.

Figures (22)

• Figure 1: Left: The color-magnitude diagram of the stars in Mutlu-Pakdil+18 are shown in gray circles. We selected objects along a 13 Gyr isochrone with $\mathrm{\,[Fe/H]}$ = $-2$, shown in black circles. The stars we observed for this work are outlined in green. Right: The fraction of stars we observed per 0.5 magnitude bin. The completeness is $\sim$ 70% for the MSTO stars.
• Figure 2: Membership selection criteria. Left: Color-magnitude diagram of Reticulum II using Mutlu-Pakdil+18 photometry. We overlaid a MIST isochrone of $\mathrm{\,[Fe/H]}$ = $-2$ and age = 13 Gyr. The red filled region is $\Delta\delta(g-r) \pm 0.08$ from the isochrone. Stars with $\Delta (g-r) > 0.08$ from the isochrone were rejected. Detection means that a metallicity was measured from a detected Ca II K line. Nondetections have 3$\sigma$ upper limits. Center: Proper motions for the subset of stars in Gaia DR3. The black dashed lines ($\mu_\alpha = 2.38$ mas yr$^{-1}$ and $\mu_\delta = -1.38$ mas yr$^{-1}$ ) are the Gaia mean proper motions of Ret II from Pace2022. Stars with proper motions $> 3\sigma$ away from that of Ret II were rejected. There were no member nondetections (open circles) with proper motions. Right: Velocity distribution of the observed sample. The member histogram includes detections and nondetections. The black dashed line is the mean velocity of Ret II $\text{v}_\text{hel} = 63.9$ km s$^{-1}$ from Ji+2023 and the red dashed line is $\text{v}_\text{hel} = 125$ km s$^{-1}$. Stars with $\text{v}_\text{hel}>$ 125 km s$^{-1}$ were rejected. The nonmembers with v $<$ 125 km s$^{-1}$ are removed by other cuts. Nonmember stars in each panel are those rejected by combining all three cuts.
• Figure 3: Example spectra with similar $B-V$ around the Calcium HK region. The red region is centered on the Ca II K line at 3933.7 Å with different bands of the KP Index. The grey regions on either side of the spectrum are used for local continuum normalization. The top two spectra have SNR $\sim$ 11.5 and the bottom two have SNR $\sim$ 4.5. The calibration uncertainty, $e_\text{calib}$, is the largest source of uncertainty and dominates 75% of our stars. The spectroscopic uncertainty, $e_\text{spec}$, decreases with higher metallicity and SNR.
• Figure 4: The equivalent width (EW) and $B-V$ color index used to get metallicity for member stars with detected Ca K lines. Metallicity increases with a larger EW and lower $B-V$. The gray contour lines are constant metallicity. The typical EW uncertainties are shown at the top in black squares for 0.1 magnitude bins. At $B-V \lesssim 0.5$, we see a clear gap between stars with $\mathrm{\,[Fe/H]}$ = $-2$ (bright green) and stars with $\mathrm{\,[Fe/H]}$ = $-2.5$ (dark green), indicating possible bimodality in $\mathrm{\,[Fe/H]}$. A discrete colorbar was used to emphasize the gap in metallicity.
• Figure 5: The color-magnitude diagram of members in Reticulum II. We included a 13 Gyr MIST isochrone with [Fe/H]= $-2$. All previous spectroscopic observations are in squares. Stars with detected Ca II K lines have measured [Fe/H] abundances in closed circles. Stars with upper limit metallicities are in open circles. This diagram emphasizes (1) how faint we were able to measure metallicities compared to what we currently have in the literature and (2) that we have observations of the main sequence for the first time in Reticulum II.