We present the results from a new search for candidate galaxies at z ≈ 8.5–11 discovered over the 850 arcmin2 area probed by the Cosmic Assembly Near-Infrared Deep Extragalactic Legacy Survey (CANDELS). We use a photometric-redshift selection including both Hubble and Spitzer Space Telescope photometry to robustly identify galaxies in this epoch at H 160 < 26.6. We use a detailed vetting procedure, including screening against persistence and stellar contamination, and the inclusion of ground-based imaging and follow-up Hubble Space Telescope imaging to build a robust sample of 11 candidate galaxies, three presented here for the first time. The inclusion of Spitzer/IRAC photometry in the selection process reduces contamination, and yields more robust redshift estimates than Hubble alone. We constrain the evolution of the rest-frame ultraviolet luminosity function via a new method of calculating the observed number densities without choosing a prior magnitude bin size. We find that the abundance at our brightest probed luminosities (M UV = − 22.3) is consistent with predictions from simulations that assume that galaxies in this epoch have gas depletion times at least as short as those in nearby starburst galaxies. Due to large Poisson and cosmic variance uncertainties, we cannot conclusively rule out either a smooth evolution of the luminosity function continued from z = 4–8, or an accelerated decline at z > 8. We calculate that the presence of seven galaxies in a single field Extended Groth Strip is an outlier at the 2σ significance level, implying the discovery of a significant over-density. These scenarios will be imminently testable to high confidence within the first year of observations of the James Webb Space Telescope.

We present a detailed stellar population analysis of 11 bright (H < 26.6) galaxies at z = 9–11 (three spectroscopically confirmed) to constrain the chemical enrichment and growth of stellar mass of early galaxies. We use the flexible Bayesian spectral energy distribution (SED) fitting code Prospector with a range of star formation histories (SFHs), a flexible dust attenuation law, and a self-consistent model of emission lines. This approach allows us to assess how different priors affect our results and how well we can break degeneracies between dust attenuation, stellar ages, metallicity, and emission lines using data that probe only the rest-frame ultraviolet (UV) to optical wavelengths. We measure a median observed UV spectral slope β = − 1.87 − 0.43 + 0.35 for relatively massive star-forming galaxies ( 9 < log ( M ⋆ / M ⊙ ) < 10 ), consistent with no change from z = 4 to z = 9–10 at these stellar masses, implying rapid enrichment. Our SED-fitting results are consistent with a star-forming main sequence with sublinear slope (0.7 ± 0.2) and specific star formation rates of 3–10 Gyr−1. However, the stellar ages and SFHs are less well constrained. Using different SFH priors, we cannot distinguish between median mass-weighted ages of ∼ 50–150 Myr, which corresponds to 50% formation redshifts of z 50 ∼ 10–12 at z ∼ 9 and is of the order of the dynamical timescales of these systems. Importantly, models with different SFH priors are able to fit the data equally well. We conclude that the current observational data cannot tightly constrain the mass-buildup timescales of these z = 9–11 galaxies, with our results consistent with SFHs implying both a shallow and steep increase in the cosmic SFR density with time at z > 10.