Title:On the redshift-evolution of the Lyman-alpha escape fraction and the dust content of galaxies

Abstract:The Lya emission line has repeatedly proven itself to be a powerful tool by which to identify and study evolving galaxies at the highest redshifts. In this context, and in order to use Lya as a probe of the physical properties of galaxies, it becomes vital to know the Lya escape fraction, fesc(Lya). Unfortunately, due to the resonant nature of Lya, fesc(Lya) may vary in an unpredictable manner and requires empirical measurement to the limits of the available data at all z. Here we compile Lya luminosity functions between z=0 and 8 from various studies and, combined with available H-alpha and UV LFs, assess how fesc(Lya) evolves with z. We find a strong upwards evolution in fesc(Lya) with z increasing between 0.3 and ~6, which is well fit by a power-law of the form fesc(Lya) \propto (1+z)^{2.6+/-0.2}. This relationship predicts that fesc(Lya) should reach 1 by z $\approx$ 11. By comparing fesc(Lya) and dust attenuation in individual galaxies we derive a new empirical relationship between fesc(Lya) and the attenuation suffered by the stellar continuum. This includes the resonance scattering physics and is able to explain the evolution of fesc(Lya) between z=0 and 6 as a function of the evolution in the overall dust content. Beyond z $\approx$ 6.5, fesc(Lya) is shown to drop substantially; an effect that can be attributed to either ionizing photon leakage, or an increase in the neutral gas content of the IGM. While distinguishing between those two scenarios may be extremely challenging, by framing the problem in this manner we remove the uncertainty of the dark matter halo mass evolution from tests of reionization based upon Lya. We finally re-arrange our expressions to derive a new method by which to estimate the dust content of galaxies at a given redshift, based purely upon the observed Lya and UV (or H-alpha) LFs. These data are well fit by an exponential function with a characteristic redshift of $\approx$ 3.5.