Evolved Massive Stars in the Local Group. I. Identification of Red Supergiants in NGC 6822, M31, and M33

Abstract

Knowledge of the red supergiant (RSG) population of nearby galaxies allows us to probe massive star evolution as a function of metallicity; however, contamination by foreground Galactic dwarfs dominates surveys for red stars in Local Group galaxies beyond the Magellanic Clouds. Model atmospheres predict that low-gravity supergiants will have B-V values that are redder by several tenths of a magnitude than foreground dwarfs at a given V-R color, a result that is largely independent of reddening. We conduct a BVR survey of several fields in the Local Group galaxies NGC 6822, M33, and M31 as well as neighboring control fields and identify RSG candidates from CCD photometry. The survey is complete to V = 20.5, corresponding to MV = -4.5 or an Mbol of -6.3 for the reddest stars. Follow-up spectroscopy at the Ca II triplet of 130 stars is used to demonstrate that our photometric criterion for identifying RSGs is highly successful (96% for stars brighter than V = 19.5; 82% for V = 19.5-20.5). Classification spectra are also obtained for a number of stars in order to calibrate color with spectral type empirically. We find that there is a marked progression in the average (B-V)0 and (V-R)0 colors of RSGs in these three galaxies, with the higher metallicity systems having a later average spectral type, which is consistent with previous findings by Elias, Frogel, & Humphreys for the Milky Way and Magellanic Clouds. More significantly, we find that there is a clear progression with metallicity in the relative number of the highest luminosity RSGs, a trend that is apparent both in absolute visual magnitude and in bolometric luminosity. Thus any use of RSGs as distance indicators requires correction for the metallicity of the parent galaxy. Our findings are in accord with the predictions of the Conti scenario in which higher metallicities result in higher mass-loss rates, resulting in a star of a given luminosity spending an increasing fraction of its He-burning lifetime as a Wolf-Rayet (WR) star rather than as an RSG with increasing metallicity. The fact that the distribution of luminosities have extended tails extending to higher luminosity (with the possible exception of the M31 sample, which contains only one high-luminosity RSG) suggests that many WRs may pass through an RSG phase, albeit for a short period. We find that over the factor of 5 difference in metallicity from NGC 6822 to M31, the masses of the most luminous RSGs change by about a factor of 2, with the highest mass RSGs having masses of 25-30 M☉ in NGC 6822, 18 M☉ in M33, and 13-15 M☉ in M31. These masses are lower than the usually assumed limits for evolution to the WR stage but are not necessarily in conflict with the number of unevolved O and B stars if the time spent as a He-burning object is split between the RSG and WR phases over a large mass range.