A homogeneous view of asymptotic giant branch carbon stars as seen by Gaia

Abstract

Carbon stars on the asymptotic giant branch are major contributors to galactic dust enrichment, with gas mass-loss rates up to 1e-4 Msun/yr. We present a homogeneous spectral energy distribution analysis of the Gaia DR3 Golden Sample of carbon stars in the Milky Way and Magellanic Clouds. Our dataset includes 14,747 sources with multi-band photometry from Gaia, 2MASS, and WISE, combined with recent distance and extinction estimates. For a subsample of 2,494 Mira variables, we model multi-band light curves to derive accurate mean magnitudes. Stellar and circumstellar parameters are obtained by fitting observations with a large grid of synthetic spectra computed with the DUSTY radiative transfer code using COMARCS atmospheres. We derive effective temperature, optical depth, and gas mass-loss rate for each source. The distributions peak around Teff = 3150 K, with mass-loss rates spanning 1e-11 to 1e-4 Msun/yr and inner dust temperatures near 1000 K. We find a correlation between variability amplitude and mass-loss rate. This framework provides a statistically robust view of carbon stars across environments with different metallicities. Apparent environmental dependencies are influenced by luminosity distributions and selection effects rather than purely intrinsic metallicity differences. The combined Gaia and WISE selection limits the detection of both highly obscured and faint Magellanic Cloud sources, but the observed trends remain significant within the sampled populations.

Figures (13)

• Figure 1: Top panel: Sky distribution of the targets in the final catalogue. Bottom panel: Extinction-corrected color-absolute magnitude diagram ($J-K_s$ vs. $M_{K_s}$). The right sub-panel shows the histograms of the absolute $K_s$ magnitude probability density function (PDF) for the three populations. The color coding is as follows: SMC stars in green, LMC in red, and MW in blue.
• Figure 2: Absolute difference between the un-reddened Gaia BP, RP and G magnitudes and our mean magnitudes derived from the light curve fit. Colours are the same as in Fig. \ref{['fig:sky-cmd']}. The y-axis range is restricted to highlight the region of interest, excluding a few outliers. The dashed black line shows the linear fit, and the shaded region denotes the 1$\rm \sigma$ confidence interval of the regression. The resulting fit parameters (slope and intercept) and their uncertainties are reported within each panel.
• Figure 3: Examples of SED fits for six stars with different stellar and dust parameters. Red points indicate the observed fluxes normalized to the stellar luminosity, the blue line shows the best-fitting model, and the small black crosses represent the model fluxes convolved with the observational filter transmission curves, which are the quantities used in the SED fitting.
• Figure 4: Distribution of $T_\mathrm{eff}$ for C-region stars in the MW, LMC and SMC. Colours and shapes of the three subsamples are showed in the legend.
• Figure 5: Distribution of mass-loss rates for stars in the MW, LMC and SMC. Colours and shapes of the three subsamples are showed in the legend.