Investigating the metallicity dependence of the mass-loss rate relation of red supergiants

TL;DR

This study probes whether the mass-loss rates of red supergiants depend on metallicity by uniformly deriving $\dot{M}$ from spectral energy distributions across the SMC, NGC 6822, the Milky Way, M31, and M33, with comparisons to the LMC. Using the radiative transfer code DUSTY under consistent assumptions, the authors quantify dust-shell properties and reveal $\dot{M}$ spanning $\sim10^{-9}$ to $10^{-5}$ $M_\odot$ yr$^{-1}$ (mean ~ $1.5\times10^{-7}$ $M_\odot$ yr$^{-1}$). They identify a kink in the $\dot{M}$–$L$ relation around $\log(L/L_\odot)\approx4.65$ in the SMC and find no strong, global metallicity trend in $\dot{M}$ across galaxies, though dust production (optical depth $\tau_V$) does increase with metallicity. The results suggest that metallicity plays a limited role in determining steady-state RSG winds, highlighting uncertainties in gas-to-dust ratios and the need for higher-quality mid-IR data (e.g., from JWST) to refine the metallicity dependence. These findings have implications for stellar evolution modeling and the role of RSG winds in galactic chemical enrichment.

Abstract

Red supergiants (RSGs) are cool and evolved massive stars exhibiting enhanced mass loss compared to their main sequence phase, affecting their evolution and fate. However, the theory of the wind-driving mechanism is not well-established and the metallicity dependence has not been determined. We aim to uniformly measure the mass-loss rates of large samples of RSGs in different galaxies with $-0.7\lesssim[Z]\lesssim0$ to investigate whether there is a potential correlation with metallicity. We collected photometry from the ultraviolet to the mid-infrared for all our RSG candidates to construct their spectral energy distribution (SED). Our final sample includes 893 RSG candidates in the Small Magellanic Cloud (SMC), 396 in NGC 6822, 527 in the Milky Way, 1425 in M31, and 1854 in M33. Each SED was modelled using the radiative transfer code DUSTY under the same assumptions to derive the mass-loss rate. The mass-loss rates range from approximately $10^{-9} \ M_{\odot}$ yr$^{-1}$ to $10^{-5} \ M_{\odot}$ yr$^{-1}$ with an average value of $1.5\times10^{-7} \ M_{\odot}$ yr$^{-1}$. We provided a new mass-loss rate relation as a function of luminosity and effective temperature for both the SMC and Milky Way and compared our mass-loss rates with those derived in the Large Magellanic Cloud (LMC). The turning point in the mass-loss rate vs. luminosity relation differs by around 0.2 dex between the LMC and SMC. The mass-loss rates of the Galactic RSGs at $\log(L/L_\odot)<4.5$ were systematically lower than those determined in the other galaxies, possibly due to uncertainties in the interstellar extinction. We found 60-70% of the RSGs to be dusty. The results for M31 and M33 are inconclusive because of source blending at distances above 0.5 Mpc, given the resolution of Spitzer. Overall, we found similar mass-loss rates among the galaxies, indicating no strong correlation with metallicity.

Figures (24)

• Figure 1: HR diagram of the RSG samples in the LMC (blue circles), SMC (orange crosses), Milky Way (black squares), and the NGC 6822 (green triangles). Four POSYDON evolutionary tracks for $[Z]=-0.7$ and $[Z]=0$ are shown in red and dotted dark-red lines, respectively, and the points indicate intervals of 10,000 yr. The upper and right panels show the density histograms (area equals to 1) of the effective temperature and luminosity, respectively.
• Figure 2: Examples of SED fits of SMC (top row), Galactic (middle), and NGC 6822 (bottom row) RSGs. The orange diamonds show the observations and the black line is the best-fit model consisting of the attenuated flux (dashed light blue), the scattered flux (dot-dashed grey) and dust emission (dotted brown). The top left corner includes the coordinates of each source in degrees. Notes: All SED fit figures are available in an electronic form via https://zenodo.org/records/16736627.
• Figure 3: Distribution of the minimum $\chi^2_\mathrm{mod}$ for the LMC (blue bar), SMC (orange bar), MW (black line), and NGC 6822 (green line). The dashed vertical line indicates the considered limit for a good fit at $\chi_\mathrm{mod}^2=1.4$.
• Figure 4: Distribution of the minimum $\tau_V$ for the RSGs in the LMC (blue bar), SMC (orange bar), MW (black line), and NGC 6822 (green dashed line). The dashed and dot-dashed histograms correspond to the RSGs in M31 and M33, respectively.
• Figure 5: Top: Mass-loss rates vs. luminosity for SMC (left) and MW (right). The colour bar shows the $T_\mathrm{eff}$. The grey triangles represent the upper limits, and the open circles indicate the RSGs with spectral classifications. The magenta points correspond to the prediction of the derived $\dot{M}(L, T_\mathrm{eff})$ relation. The dashed blue lines show the prescription from Beasor_2020 for initial masses 10 and 20 $M_\odot$ and the orange dot-dashed line is the one from deJager_1988. Middle: MAD of $W1$ band vs. luminosity diagram for the SMC (left) and the MW (right). The colour bar shows the mass-loss rate. Bottom: Mass-loss rates vs. the effective temperature for the SMC (left) and the MW (right) indicating the luminosity with colour.