Measuring Star Formation Histories from Asymptotic Giant Branch Stars I: A Demonstration in M31

TL;DR

This work introduces a method to derive quantitative star formation histories by fitting near-infrared CMDs of resolved AGB stars with COLIBRI TP-AGB isochrones within the MATCH CMD framework. The approach is validated by reproducing the outer-disk SFH of M31 as measured from optical CMDs and is then applied to six regions in M31's inner halo to produce spatially resolved SFHs, revealing a dominant 3–5 gigayear-ago star formation episode and a total stellar mass formed over the past several gigayears. A key result is that ~1000 AGB stars suffice for robust SFH recovery, owing to the strong age sensitivity of AGB stars in the NIR, and the method yields SFHs in agreement with optical-based measurements despite using far fewer stars. The study highlights the potential of resolved AGB stars to trace SFHs back to ancient epochs with future infrared facilities, enabling measurements out to tens of megaparsecs with JWST, Roman, and Euclid, and it discusses current caveats and avenues for reducing systematic uncertainties via improved AGB models and broader data sets.

Abstract

We demonstrate how near-infrared (NIR) imaging of resolved luminous asymptotic giant branch (AGB) stars can be used to measure well-constrained star formation histories (SFHs) across cosmic time. Using UKIRT $J$ and $K$-band imaging of M31, we first show excellent agreement over the past $\sim8$ Gyr between the PHAT SFH of M31's outer disk derived from a deep optical color-magnitude diagram (CMD; $\sim3.3\times10^{7}$ stars with $M_{\rm F814W} \lesssim +2$), and our spatially-matched SFH based only on modeling AGB stars on a NIR CMD ($\sim7.7\times10^{3}$ stars with $M_{\rm J} \lesssim -6$). We find that only $\sim1000$ AGB stars are needed for reliable SFH recovery, owing to their excellent age sensitivity in the NIR. We then measure the spatially-resolved SFH of M31's inner stellar halo ($D_{\rm M31, projected} \sim20-30$ kpc) using $\sim10^4$ AGB stars. We find: (i) a dominant burst of star formation across M31's stellar halo $3-5$ Gyr ago and lower level, spatially distributed star formation $\sim1-3$ Gyr ago; (ii) $M_{\star}\approx3_{-1}^{+5}\times10^9 M_{\odot}$ formed over the past $\sim8$ Gyr. We discuss some caveats and the enormous potential of resolved AGB stars in the NIR for measuring SFHs back to ancient epochs ($\sim14$ Gyr ago) in galaxies to large distances ($D\gtrsim20$ Mpc) with JWST, Roman, and Euclid.

Figures (21)

• Figure 1: Hertzsprung-Russel diagram for four different PARSEC-COLIBRI isochrones with metallicity [M/H] = 0.0 dex. The TP-AGB evolutionary stage is shown in red, the E-AGB phase is shown in blue, and the RGB phase is shown in orange. The red helium-burning (RHeB) evolutionary phase is also labeled.
• Figure 2: Simulated CMDs using PARSEC-COLIBRI models that demonstrate the power of AGB stars (highlighted within the grey box) as age indicators from $\sim50$ Myr to $\sim10$ Gyrs. All CMDs were constructed using a constant star formation rate (SFR) of 0.1 $M_{\odot}~\rm{yr^{-1}}$ with a constant metallicity of [M/H] = 0.0 dex. The legend shows the log(age) ranges of each SSP.
• Figure 3: COLIBRI isochrones of the asymptotic giant branch with a constant age of 1 Gyr and a range of metallicities. Metallicity has a modest affect on AGB star positions in a NIR CMD compared to age.
• Figure 4: Comparisons between our 'injected' star formation histories and 'recovered' star formation histories in the two mock galaxies described in § \ref{['sec:mock_summary']}. (Left) Density maps for the observed, best-fit, and residual CMDs, expressed in units of Poisson standard deviations. (Middle) Cumulative stellar mass formed as a function of time. (Right) Total star formation rate as a function of time. These tests illustrate our ability to self-consistently recover SFHs from AGB stars.
• Figure 5: (Left) CMD of the M31 PHAT optical data analyzed in williams17 for $d>11$ kpc. (Right) CMD of our M31 UKIRT data that overlaps with the PHAT footprint analyzed in williams17 for $d>11$ kpc. Color indicates the number of stars in each CMD bin. The grey box shows our selection criteria for the AGB stars, which also corresponds to the grey box in Figure \ref{['fig:isochrones']}, shifted to M31's distance. In both CMDs, isochrones of 100 Myr, 1 Gyr, and 8 Gyr with a metallicity of [M/H]=0.0 dex are overlaid with the TP-AGB evolutionary phase highlighted in red. The PHAT optical CMD contains $\sim47$ million stars of which $\sim33$ million stars are above the 50% completeness limit ($F814W\approx 26.5$ mag). Our NIR CMD contains $\sim 35,000$ stars and an average 50% completeness limit of $J = 18.7$ mag. Our selection region contains $\sim 7,700$ AGB stars from which we measure our SFH. Typical CMD features, such as BHeB (Blue Helium-Burning) stars, RHeB (Red Helium-Burning) stars, AGB stars, RGB stars, MS (Main Sequence) stars, RC stars, and MW foreground stars are labeled.