Constraints on the Ionizing Efficiency of the First Galaxies

Observations of the Lyman-alpha forest and of high-redshift galaxies at z~5-10 imply that there were just enough photons to maintain the universe in an ionized state at z~5-6, indicating a "photon-starved" end to reionization. The ionizing emissivity must have been larger at earlier times in order to yield the extended reionization history implied by the electron scattering optical depth constraint from WMAP. Here we address the possibility that a faint population of galaxies with host halo masses of ~1e8-1e9 Msun dominated the ionizing photon budget at redshifts of about z>9, due to their much higher escape fractions. Such faint, early galaxies, would not have formed in ionized regions due to suppression by heating from the UV background (UVB), and would therefore not contribute to the ionizing background at z<6, after reionization is complete. Our model matches: (1) the low escape fractions observed for high-redshift galaxies, (2) the WMAP constraint of tau_es~0.09, (3) the low values for the UVB at z<6, and (4) the observed star formation rate density inferred from Lyman-break galaxies. A top heavy IMF from Pop III stars is not required in this scenario. We compare our model to recent ones in the literature that were forced to introduce an escape fraction that increases strongly towards high redshift, and show that a similar evolution occurs naturally if low mass galaxies possess high escape fractions.