The Hubble Space Telescope Advanced Camera for Surveys Emission Line Survey of Andromeda. I: Classical Be Stars

TL;DR

This study uses Hubble Space Telescope data to analyze classical Be stars in Andromeda, revealing their distribution, evolution, and comparison with other galaxies, and providing insights into their transient disk phenomena.

Contribution

First detailed HST survey of Be stars in Andromeda, quantifying their fraction, spectral distribution, and transient disk behavior in a metal-rich environment.

Findings

Be stars constitute about 6% of B-type stars in Andromeda.

The Be fraction decreases with spectral subtype and metallicity.

Disk-loss and regeneration episodes occur at a rate of approximately 22% per year.

Abstract

We present results from a 2-epoch HST H$\alpha$ emission line survey of the Andromeda Galaxy that overlaps the footprint of the Panchromatic Hubble Andromeda Treasury (PHAT) survey. We find 552 (542) classical Be stars and 8429 (8556) normal B-type stars in epoch # 1 (epoch # 2), yielding an overall fractional Be content of 6.15% $\pm$0.26% (5.96% $\pm$0.25%). The fractional Be content decreased with spectral sub-type from $\sim$23.6% $\pm$2.0% ($\sim$23.9% $\pm$2.0%) for B0-type stars to $\sim$3.1% $\pm$0.34% ($\sim$3.4% $\pm$0.35%) for B8-type stars in epoch # 1 (epoch # 2). We observe a clear population of cluster Be stars at early fractional main sequence lifetimes, indicating that a subset of Be stars emerge onto the ZAMS as rapid rotators. Be stars are 2.8x rarer in M31 for the earliest sub-types compared to the SMC, confirming that the fractional Be content decreases in significantly more metal rich environments (like the Milky Way and M31). However, M31 does not follow a clear trend of Be fraction decreasing with metallicity compared to the Milky Way, which may reflect that the Be phenomenon is enhanced with evolutionary age. The rate of disk-loss or disk-regeneration episodes we observe, 22% $\pm$ 2% yr$^{-1}$, is similar to that observed for seven other Galactic clusters reported in the literature, assuming these latter transient fractions scale by a linear rate. The similar number of disk-loss events (57) as disk-renewal events (43) was unexpected since disk dissipation time-scales can be $\sim$2x the typical time-scales for disk build-up phases.