How invisible stellar halos bias our understanding of ultra-faint galaxies

Abstract

We explore how a realistic surface brightness detection limit of $μ_V \approx 32.5$ mag arcsec$^{-2}$ for stars at the edges of ultra-faint galaxies affects our ability to infer their underlying properties. We use a sample of 19 galaxies with stellar masses $\approx 400 - 40,000~{\rm M}_\odot$ simulated with FIRE-2 physics and baryonic mass resolution of $30~M_{\odot}$. The surface brightness cut leads to smaller sizes, lower stellar masses, and lower stellar velocity dispersions than the values inferred without the cut. However, by imposing this realistic limit, our inferred galaxy properties lie closer to observed populations in the mass-size plane, better match observed velocity dispersions as a function of stellar mass, and better reproduce derived circular velocities as a function of half-light radius. For the most massive galaxies in our sample, the surface brightness cut leads to higher mean $\rm [Fe/H]$ values, but the increase is not enough to match the observed MZR. Finally, we demonstrate that the common Wolf et al. (2010) mass estimator is less accurate when the surface brightness cut is applied. For our lowest-mass galaxies, in particular, excluding the low-surface brightness outskirts causes us to overestimate their central dark-matter densities and virial masses. This suggests that attempts to use mass estimates of ultra-faint galaxies to constrain dark-matter physics or to place constraints on the low-mass threshold of galaxy formation must take into account surface brightness limits or risk significant biases.