As extensively shown in the previous section, the Lyα emission is very sensitive to even a tiny fraction of HI along the line of sight. In this regard, studying the LAE population and the evolution of its detected number counts with redshift, or line of sight can give constraints on zeor and its
variations. LAEs do have the same spectral signature as quasars, but since they are not as bright, they are less suited for detailed spectra analysis and are in general used to measure statistical properties.
LAE fraction. The observed fraction of LAEs among UV selected galaxies is the simplest probe of end of reionization that can be used with LAEs. For z > zeor, it is expected that the number of
detected LAEs drops significantly since most of Lyα emission would be absorbed. By extrapolating the trend observed at z < zeorto z > zeor, and comparing this to the actual measurements obtained
at z > zeor, it is possible to estimate ¯xHI. Using this technique, Caruana et al. (2014) reported
¯xHI ∼ 0.5 at z ∼ 7 and Pentericci et al. (2014) concluded that ¯xHI ≥ 0.51 for z ∼ 7 was needed to
explain the observations, assuming that the evolution was only caused by an evolving IGM. More recently, more precise measurements of ¯xHI= 0.59+0.11−0.15at z ∼ 7 and ¯xHI≥ 0.76 at z ∼ 8 (68% lower
limit) were obtained in respectively Mason et al. (2018) and Mason et al. (2019), through improved modeling of the IGM and the use of reionization simulations. These results are consistent between themselves, and it can be noted that the average estimate of ¯xHI derived from galaxy observation
in favor of the fact that quasar do tend to probe over dense regions of the IGM where ionizing fronts progress faster.
Escape fraction. Another approach is to study the evolution Lyα photons escape fraction. Under normal star forming condition, fLyα can be computed by comparing the SFR obtained from measurments of the dust corrected monochromatic UV continuum (L1500) to the one estimated from
LLyα. The idea behind that, is that using L1500we can recover the intrinsic SFR and assuming that
the Lyα emission is only powered by star forming regions (i.e., no AGN), the observed discrepancy between SFRUV and SFRLyα can only be explained by an increased absorption of Lyα photons.
Using a Hα to Lyα line ratio of 8.7, assuming a case B recombination (Osterbrock & Ferland, 2006), a Salpeter IMF and the prescription for SFR computation in Kennicutt, Jr. (1998), the following formula can be used to estimate fLyα:
fLyα = SFRLyα SFRUV = 7.9 × 10 42LLyα 8.7 × 1 10−28L1500 ∝ LLyα L1500 (1.3) where LLyα and L1500 are expressed in respectively erg s−1 and erg Hz−1s−1. This simple prescrip-
tion does not allow to differentiate between the effect of dust, neutral hydrogen in the ISM of star forming regions, or neutral hydrogen present in the IGM. In addition, it is only valid assuming a stationary regime of star formation has been established for longer than ∼ 100Myr. In a transition regime of star formation, the typical timescale of escape of a Lyα photon being much longer (possi- bly several millions of years) than the one of a UV photon, the SFR measured from LLyαis lagging
behind the SFR measured from L1500 which provides a better “instantaneous” measurement.
Observations have shown that fLyα is increasing from z = 0 to z ∼ 6, and drops for z & 6. This can be explained by a lower dust content at higher redshift and an increasing amount of neutral hydrogen at z > 6 (see e.g., Ono et al., 2010; Blanc et al., 2011; Hayes et al., 2011) which once again, places zeor around 6. Since it is not possible to differentiate between the effects of dust,
ISM and IGM in the observation of fLyα, study of lower redshift galaxies are needed to disentangle between these effects.
A good introduction to this approach is provided in Matthee et al. (2016); Sobral et al. (2017); Sobral & Matthee (2019) presenting the CALYMHA program (CAlibrating LYMan-α with Hα), a survey of Hα selected galaxies at z ∼ 2 designed to investigate the impact of galaxy properties on observed Lyα related quantities such as flux, EWLyα or fLyα. It has been observed in Matthee et al. (2016) that fLyα tends to be anti correlated with SFR, and that while it is naively expected that high mass galaxies would have lower fLyα due to higher dust content, this is not seen in this work. These two observations suggest that the kinematics of the gas and dust content may have a greater impact on fLyα than previously thought, and that therefore fLyα can show great variations from one galaxy to another.
The translation of the conclusions developed in these studies to higher redshift and the study of reionization is non trivial, and some progress are yet to be made on the understanding of the selection bias affecting such work. However, this offers a promising line of work to understand and estimate the escape fraction of high redshift galaxies. As illustration of this statement, it has been shown that the strong correlation that exists between fLyα and EWLyα could be used to estimate fLyα Verhamme et al. (2017); Sobral & Matthee (2019). In addition, this correlation does not show significant evolution over the redshift range z = 0 − 5 (Harikane et al., 2018), which justifies the study of low redshift analogs to understand the physics of high redshift galaxies.
LAE clustering. At the very beginning of reionization, only the brightest galaxies could create an HII region large enough for their Lyα photons to be redshifted out of resonance before reaching the
neutral medium (see the "toy model” presented in Matthee et al., 2015). As reionization made its progress, these HII regions expanded, progressively unveiling more LAEs within the denser regions. In terms of observations this means that for z & zeor, LAEs appear more clustered compared to UV
selected galaxies (see e.g. Orsi et al., 2008; Sobacchi & Mesinger, 2015; Ouchi et al., 2018). The fact that ionizing fronts progress faster along over-dense regions has not always been the general understanding since these regions also attract more neutral gas in their surrounding, which has the opposite effect.
As such, the clustering effect can be used to probe and quantify the inhomogeneity of the end of reionization, and can be compared to the different measurements of zeor (see e.g., the slight
differences between zeor measured from quasar spectra and zeor derived from the evolution of LAE
fraction mentioned at the beginning of this section). But the measurement of this clustering can also be compared to theoretical predictions and used to constrain ¯xHI. This was done for example,
in Ouchi et al. (2018) who measured ¯xHI = 0.15 ± 0.15 using the clustering signal of LAEs at
z ∼ 6.6.
LAE spectral profile. The last probe of end of reionization with LAEs is the spectral profile of the Lyα line itself. Hu et al. (2016) reported the discovery of a double peak LAE at z = 6.59, which remains at this time, the only double peaked LAE at z > 6. This observation was unexpected since the transmission of the blue part of the Lyα emission is often very low (Dijkstra et al., 2007; Laursen et al., 2011), especially at high redshift where the Lyα radiation is easily absorbed by the surrounding neutral hydrogen.
The original publication states that the detection of this blue wing could be a indicative of either a highly ionized region created by an extremely UV bright source (as suggested by the model in Matthee et al. (2015)), or a high velocity shift with respect to the IGM. The improved analysis in Matthee et al. (2018) confirms this discovery and investigates various possible scenarios to explain this observed profile. The conclusion is that the large velocity offset scenario seems to be unlikely, leaving the large HII region as a reasonable explanation for this observed profile.
Since this galaxy is not identified as an AGN which typically have broader Lyα lines, this could be one of the first direct evidence of a star forming galaxy actively ionizing its surrounding medium. This discovery potentially opens the door to a new approach for the study of reionization if more objects like this are found in the future.