Investigating episodic mass loss in evolved massive stars IV. Comprehensive analysis of dusty red supergiants in NGC 6822, IC 10, and WLM

TL;DR

Massive stars lose mass during the red supergiant phase, but the mechanisms are not well understood, especially at low metallicity. This study analyzes seven dusty RSGs in NGC 6822, IC 10, and WLM using near-IR J-band spectroscopy from EMIR (GTC) with NLTE-corrected MARCS atmospheres, optical reclassifications, and multi-band time-series photometry to search for episodic mass-loss signatures. The authors derive $T_{ m eff}$, $ m log g$, [Fe/H], $\xi$, and $v_{ m rad}$ via spectral fitting, compare Teff with independent relations, and construct light curves to identify variability and periods; they identify four RSGs with evidence for episodic mass loss and report a candidate-dimming event in NGC 6822-175. The results demonstrate that episodic mass loss exists in dusty, low-metallicity RSGs and that combining spectroscopy with long-baseline photometry is crucial for diagnosing these events, informing mass-loss physics and late-stage stellar evolution in galaxies beyond the Milky Way, and guiding future JWST and LSST studies.

Abstract

Mass loss shapes the fate of massive stars; however, the physical mechanism causing it remains uncertain. We present a comprehensive analysis of seven red supergiants, for which we searched evidence of episodic mass loss, in three low-metallicity galaxies: NGC~6822, IC~10, and WLM. Initially, the spectral classification of their optical spectra was refined and compared to previous reported classifications, finding four sources that display spectral variability. We derived the physical properties of five of them using the \textsc{marcs} atmospheric models corrected for nonlocal thermal equilibrium effects to measure stellar properties from our new near-infrared spectra, such as the effective temperature, surface gravity, metallicity, and microturbulent velocity. Additional empirical and theoretical methods were employed to calculate effective temperatures, finding consistent results. We constructed optical and infrared light curves, discovering two targets in NGC~6822 with photometric variability between 1 and 2.5 mag in amplitude in r and ~ 0.5 mag in the mid-infrared. Furthermore, we discovered a candidate-dimming event in one of these sources. Periods for three red supergiants were determined using epoch photometry, which were consistent with the empirical estimations from literature period-luminosity relations. Our comprehensive analysis of all the available data for each target provides evidence for episodic mass loss in four red supergiants.

Figures (24)

• Figure 1: Mid-IR (left) and near-IR (right) CMDs for the seven targets in NGC 6822, IC 10 and WLM. Red circles indicate targets whose spectra we were able to model. Background sources are stars from NGC 6822. The vertical line in the mid-IR CMD indicates the color criterion used for selecting stars with IR-excess.
• Figure 2: Spatial distribution of our targets in NGC 6822 and the targets of Patrick2015, two of which we have also observed. We included RSG candidates from the same work, one of which we have observed (see text for details). The background image is from the DSS2 survey.
• Figure 3: Revized spectral types of OSIRIS targets from deWit2025, ordered in sequence of spectral type. Important spectral features are indicated. The spectrum and classification for WLM 14 are from Britavskiy2019.
• Figure 4: Light curve of NGC6822-52. Filters and surveys are specified in the legend. Vertical lines indicate the epochs spectroscopy was obtained for this source. The dash-dotted line represents the epoch of our EMIR observations, while the dotted line denotes the epoch of the OSIRIS data deWit2025.
• Figure 5: Observed spectrum of NGC6822-103 (in black) and best-fitting model (in red). The shaded blue regions highlight the spectral windows considered for the fitting process. The lines used for the fitting process from left to right by species are Fe i$\lambda$$\lambda$ 11882.847, 11973.050; Ti i$\lambda$$\lambda$ 11892.878, 11949.542; Si i$\lambda$$\lambda$ 11984.20, 11991.57, 12031.50, 12103.54, 12270.50; Mg i$\lambda$$\lambda$ 11828.185, 12083.346.