The first (nearly) model-independent constraint on the neutral hydrogen fraction at z~5--6

TL;DR

This paper provides the first model-independent upper limits on the neutral hydrogen fraction at redshifts 5 to 6, using dark pixel statistics in quasar spectra, challenging previous model-dependent constraints.

Contribution

It introduces a robust, assumption-free method to constrain the neutral hydrogen fraction at high redshift using dark pixel covering fractions in quasar absorption spectra.

Findings

<x_HI> < 0.2 at 5 < z < 5.5

<x_HI> < 0.8 at z~6.1

Tentative evidence for a change in dark pixel covering fraction at z~5.5

Abstract

Cosmic reionization is expected to be complex, extended and very inhomogeneous. Existing constraints at z~6 on the volume-averaged neutral hydrogen fraction, <x_HI>, are highly model-dependent and controversial. Constraints at z<6, suggesting that the Universe is highly ionized, are also model-dependent, but more fundamentally are invalid in the context of inhomogeneous reionization. As such, it has recently been pointed out that there is no conclusive evidence that reionization has completed by z~5--6, a fact that has important ramifications on the interpretation of high-redshift observations and theoretical models. We present the first direct upper limits on <x_HI> at z~5--6 using the simple and robust statistic of the covering fraction of dark pixels in the Ly alpha/beta forests of high redshift quasars. With a sample of 13 Keck ESI spectra we constrain <x_HI> < 0.2 at 5 < z < 5.5, rising to <x_HI> < 0.8 at z~6.1. We also find tentative evidence for a break in the redshift evolution of the dark covering fraction at z~5.5. A subsample of two deep spectra provides a more stringent constraint of <x_HI>(z=6.1) < 0.5 when combined with conservative estimates of cosmic variance. This upper limit is comparable to existing results at z~6 but is more robust. The results presented here do not rely on assumptions about the quasar continuum, IGM density, HII morphology or ionizing background fields, and thus are a good starting point for future interpretation of high redshift observations.