Observational constraints on Cosmic Reionization

TL;DR

Observational constraints on Cosmic Reionization synthesizes multi-wavelength evidence to map the EoR, contrasting Gunn-Peterson absorption, Ly$\alpha$-galaxy statistics, metal lines, GRB probes, CMB polarization, and 21 cm prospects. It argues that reionization is a extended, inhomogeneous process beginning as early as $z\sim 14$ and ending near $z\sim 6$, with large sightline-to-sightline variation near overlap. The work highlights that low-luminosity star-forming galaxies likely provide the dominant ionizing photons and that the CMB polarization and upcoming 21 cm tomography will jointly reconstruct the reionization history and topology. Overall, the paper emphasizes a transition from a simple late or early single-event picture to a nuanced, gradual reionization scenario accessible through next-generation observatories.

Abstract

Recent observations have set the first constraints on the epoch of reionization (EoR), corresponding to the formation epoch of the first luminous objects. Studies of Gunn-Peterson (GP) absorption, and related phenomena, suggest a qualitative change in the state of the intergalactic medium (IGM) at $z \sim 6$, indicating a rapid increase in the neutral fraction of the IGM, from $x_{HI} < 10^{-4}$ at $z \le 5.5$, to $x_{HI} > 10^{-3}$, perhaps up to 0.1, at $z \ge 6$. Conversely, transmission spikes in the GP trough, and the evolution of the \lya galaxy luminosity function indicate $x_{HI} < 0.5$ at $z\sim 6.5$, while the large scale polarization of the cosmic microwave background (CMB) implies a significant ionization fraction extending to higher redshifts, $z \sim 11 \pm 3$. The results suggest that reionization is less an event than a process, with the process beginning as early as $z \sim 14$, and with the 'percolation', or 'overlap' phase ending at $z \sim 6$. The data are consistent with low luminosity star forming galaxies as being the dominant sources of reionizing photons. Low frequency radio telescopes currently under construction should be able to make the first direct measurements of HI 21cm emission from the neutral IGM during the EoR, and upcoming measurements of secondary CMB temperature anisotropy will provide fine details of the dynamics of the reionized IGM.

Figures (15)

• Figure 1: Moderate resolution spectra of nineteen SDSS quasars at $5.74 < z < 6.42$. Adapted from Fan et al. (2006b)
• Figure 2: Transmitted flux blueward of Ly$\alpha$ emission as a function of redshift from $z\sim 2$ to 6.3 using Keck/ESI and HIRES data. Large open squares are points data for possible BAL quasars. Flux was computed in bins of 15Å. Adapted from Songaila (2004)
• Figure 3: Evolution of Optical depth with combined Ly$\alpha$ and Ly$\beta$ results. The dash line is for a redshift evolution of $\tau_{\rm GP} \propto (1+z)^{4.3}$. At $z>5.5$, the best fit evolution has $\tau_{\rm GP} \propto (1+z)^{>10.9}$, indicating an accelerated evolution. The large open symbols with error bars are the average and standard deviation of optical depth at each redshift. The sample variance increases also increases rapidly with redshift. Adapted from Fan et al. (2006b).
• Figure 4: Close-up of the Ly$\alpha$ and Ly$\beta$ troughs in the two highest redshift quasars currently known. SDSS J1030+0524 ($z=6.28$) shows complete Gunn-Peterson absorption in both Ly$\alpha$ and Ly$\beta$ , where SDSS J1148+5251 ($z=6.42$) has clear transmission, suggesting large line of sight variance at the end of reionization. Adapted from White et al. (2003).
• Figure 5: Distributions of dark gaps, defined as regions in the spectra where all pixel having observed optical depth larger than 2.5 for Ly$\alpha$ transition. Upward arrows are gaps immediately blueward of quasar proximity zone, therefore the length is only a lower limit. Solid lines with error bars are average depth lengths with 1-$\sigma$ dispersion at each redshift bin. Long dark gaps start to appear at $z\sim 5.6$, with the average gap length increases rapidly at $z>6$, marking the end of reionization. It is compared with simulation of Paschos & Norman (2005), in which dashed and dotted lines are for moderate and high spectral resolutions. The simulation has an overlapping redshift at $z \sim 7$. Adapted from Fan et al. (2006b).