Calibrating Eruptive Mass Loss in Red Supergiants with Local Group Data

Abstract

We calibrate a physically motivated, super-Eddington eruptive mass-loss prescription for red supergiants (RSGs) using Local Group stellar populations. Building on MESA models that add eruptive mass loss with a free scaling parameter $ξ$, we generate stellar evolution tracks and isochrones, and synthesize mock populations at metallicities of $Z/Z_\odot=0.2,\ 0.4$, and $1.0$. We compare model luminosity functions to observations of RSGs in the SMC, LMC, and M31, restricting to $3.5<\log T_{\rm eff}/K<3.75$ and $\log(L/L_\odot)>4.5$. By-eye fits to the observations yield values of $ξ_\mathrm{SMC}=0.0-0.05$, $ξ_\mathrm{LMC}=0.1$, and $ξ_\mathrm{M31}=0.35$, implying a positive, linear trend between the strength of eruptive mass-loss and metallicity. This calibrated eruptive mass loss prevents stars with initial masses $\gtrsim 20~M_\odot$ from evolving to become red supergiants, with implications for the mass spectrum of core-collapse progenitors, compact remnants, early supernova interaction signatures, and the spectral energy distributions of unresolved galaxies.

Figures (6)

• Figure 1: Eruptive mass loss rates across the H-R diagram for $Z=0.2~Z_\odot$. Left: $\xi=0.1$. Right: $\xi=0.5$. Grey lines show the evolutionary tracks of stars from the beginning of main sequence until the end of the model run. Models were run at every $1~M_\odot$ increments but only selected models are shown here for clarity. Colored contours show the rate of eruptive mass loss experienced by the star, with legend on the right.
• Figure 2: Eruptive mass loss rates across the H-R diagram for $Z=0.4~Z_\odot$. Left: $\xi=0.1$. Right: $\xi=0.5$. All axes, labels, and other lines follow Figure \ref{['fig:SMC']}.
• Figure 3: Eruptive mass loss rates across the H-R diagram for $Z=1.0~Z_\odot$. Left: $\xi=0.1$. Right: $\xi=0.5$. All axes, labels, and other lines follow Figure \ref{['fig:SMC']}.
• Figure 4: H-R diagram for $Z = 1.0~Z_\odot$. Left: $\xi=0.0$. Right: $\xi=0.35$. MESA tracks are in red lines and ArtPop-simulated population generated from the MESA model grid are in black dots. For clarity, only models from $6-50~M_\odot$ are shown. Gaussian noise of $0.1~\mathrm{dex}$ and $0.03~\mathrm{dex}$ were added for $\log(\mathrm{L/L_\odot})$ and $\log(\mathrm{T_{\rm eff}/K})$ respectively to mimic observational scatter.
• Figure 5: Cumulative luminosity functions for each modeled stellar population compared to observations of the SMC (upper left panel, $Z=0.2~Z_\odot$), LMC (upper right panel, $Z=0.4~Z_\odot$), and M31 ($Z=1.0~Z_\odot$, lower panel). Shown are both cumulative luminosity functions from ArtPop population synthesis output (colored lines) and observations (black lines). The thicker colored line (either dashed or solid) indicates the best fit model. Grey shading is the Poisson error of the observed data.