Star-forming galaxies at z approximate to 8-9 from Hubble Space Telescope/WFC3: implications for reionization

We present a search for galaxies at 7.6 < z < 9.8 using the latest Hubble Space Telescope/Wide Field Camera 3 (WFC3) near-infrared data, based on the Lyman-break technique. We search for galaxies which have large (Y - J) colours (the 'Y-drops') on account of the Lyman alpha forest absorption, and with (J - H) colours inconsistent with being low-redshift contaminants. We identify 24 candidates at redshift z approximate to 8-9 (15 are robust and a further nine more marginal but consistent with being high redshift) over an area of approximate to 50 arcmin(2). Previous searches for Y-drops with WFC3 have focused only on the Hubble Ultra Deep Field, and our larger survey (involving two other nearby deep fields and a wider area survey) has trebled the number of robust Y-drop candidates. For the first time, we have sufficient z approximate to 8-9 galaxies to fit both phi* and M* of the UV Schechter luminosity function. There is evidence for evolution in this luminosity function from z = 6-7 to z = 8-9, in the sense that there are fewer UV-bright galaxies at z approximate to 8-9, consistent with an evolution mainly in M*. The candidate z approximate to 8-9 galaxies we detect have insufficient ionizing flux to reionize the Universe, and it is probable that galaxies below our detection limit provide a significant UV contribution. The faint-end slope, alpha, is not well constrained. However, adopting a similar faint-end slope to that determined at z = 3-6 (alpha = -1.7) and a Salpeter initial mass function (IMF), then the ionizing photon budget still falls short if f(esc) < 0.5, even integrating down to M-UV = -8. A steeper faint-end slope or a low-metallicity population (or a top-heavy IMF) might still provide sufficient photons for star-forming galaxies to reionize the Universe, but confirmation of this might have to await the James Webb Space Telescope.