Cosmic reionization by stellar sources: Population II stars

We study the reionization of the Universe by stellar sources using a numerical approach that combines fast 3D radiative transfer calculations with high-resolution hydrodynamical simulations. By supplementing a one-step radiative transfer code specifically designed for following ionization processes with an adaptive ray-tracing algorithm, we are able to speed up the calculations significantly to the point where handling a vast number of sources becomes technically feasible. This allows us to study how dim low-mass sources, excluded in previous investigations owing to computational limitations, affect the morphological evolution of the reionization process. Ionizing fluxes for the sources are derived from intrinsic star formation rates computed in the underlying hydrodynamical simulations. Analysis of numerically converged results for star formation rates and halo mass functions allows us to assess the consequences of not including low-mass objects and enables us to correct for resolution effects. With these corrections, we are able to reduce the effective mass resolution limit for sources to M similar to 4.0 x 10(7) h(-1) M-circle dot in a 10 h(-1) Mpc comoving box. Our calculations reveal that the process by which ionized regions in the intergalactic medium (IGM) percolate is complex and is especially sensitive to the inclusion of dim sources. Moreover, we find that, given the same level of cosmic star formation, the number of ionizing photons required to reionize the Universe is significantly overestimated if sources with masses below similar to10(9) h(-1) M-circle dot are excluded. This result stems from the fact that low-mass sources preferentially reside in less clumpy environments than their massive counterparts. Consequently, their exclusion has the net effect of concentrating more of the cosmic ionizing radiation in regions which have higher recombination rates. We present the results of our reionization simulation assuming a range of escape fractions for ionizing photons and make statistical comparisons with observational constraints on the neutral fraction of hydrogen at z similar to 6 derived from the z = 6.28 Sloan Digital Sky Survey (SDSS) quasar of Becker and coworkers. We find that, given the amplitude and form of the underlying star formation predictions, an escape fraction near f(esc) = 0.10-0.20 is most consistent with the observational results. In these models, reionization is expected to have occurred between z similar to 7-8, although the IGM remains fairly opaque until z similar or equal to 6. Our method is also capable of handling the simultaneous reionization of the helium component in the IGM, allowing us to explore the plausibility of the scenario where sources with harder spectra are primarily responsible for reionization. In this case, we find that if the sources responsible for reionizing hydrogen by z similar to 8 had spectra similar to active galactic nuclei, then the helium component of the IGM should have been reionized by z similar to 6. We find that such an early reionization epoch for helium does not necessarily conflict with observational constraints obtained at z similar to 3, but may be challenged by future observations at higher redshifts. The recent WMAP measurements of the electron scattering optical depth (tau(e) = 0.17 +/- 0.04 according to the 'model independent' analysis of Kogut et al.) appear to be inconsistent with the relatively late onset of reionization by the normal Population II type stars that we consider. In order to simultaneously match the observations from the z = 6.28 SDSS quasar and the optical depth measurement from WMAP with the sources modelled here, we require a boosting factor for the number of ionizing photons released in the f(esc) = 0.20 model which evolves from unity at z = 6 to greater than or similar to50 by z similar to 18. Such a steep enhancement in the stellar production rate of ionizing photons would be consistent with an IMF that becomes more and more top heavy with increasing redshift.