Onset of Cosmic Reionization: Evidence of an Ionized Bubble Merely 680 Myr after the Big Bang

Cosmological simulations indicate that the process of reionization is expected to be spatially inhomogeneous (e.g., Furlanetto et al. 2004; Iliev et al. 2006; Zahn et al. 2007; Mesinger & Furlanetto 2008; Jensen et al. 2014), and started when intense UV radiation from individual galaxies or groups of galaxies first ionized their local surroundings and formed ionized bubbles of size $\simeq 1\,\mathrm{Mpc}$, which later grew to fill the entire intergalactic medium (IGM), marking the end of the reionization process (Shin et al. 2008).

Star-forming galaxies at high redshifts are expected to have contributed to the reionization process (e.g., Bouwens et al. 2015; Finkelstein et al. 2015; Dayal & Ferrara 2018). These same galaxies, via their Lyα emission, provide a practical tool to study the reionization process (e.g., Malhotra & Rhoads 2004). Overdensities of Lyα galaxies have been discovered at $z\sim 7$ (Zheng et al. 2017; Castellano et al. 2018), which may indicate large ionized regions. Overdensities of bright Lyman-break galaxies have also been found at these redshifts (e.g., Trenti et al. 2012; Castellano et al. 2016; Harikane et al. 2019). We have carried out a unique narrowband (NB) imaging survey, the Cosmic Deep And Wide Narrowband survey (DAWN; PI: Rhoads), with enough sensitivity to detect Lyα emitting galaxies at a redshift of z = 7.7 and enough area coverage to detect the possible inhomogenieties and ionized bubbles expected at early stages of reionization. The NB technique has been proven successful at identifying $z\gt 6$ Lyα galaxies (Hu et al. 2010, 2019; Ouchi et al. 2010; Rhoads et al. 2012). Here we present a discovery of the most distant galaxy group (hereafter EGS77), identified using NB imaging, and confirmed via spectroscopic observations. The visibility of faint Lyα emission indicates an ionized IGM in the local vicinity of this group.

2.1. Observations and Data Reduction

2.2. Selection of z = 7.7 Candidate Galaxies

To generate the source catalog we first aligned all images that include F606W, F814W, near-IR NB1066 (NB from DAWN survey), and near-IR broadband F125W, F160W images onto a common world coordinate system grid. We then used a source detection software (SExtractor; Bertin & Arnouts 1996) in dual image mode, where the detection image (in this case the NB) is used to identify the pixels associated with each object, while the fluxes are measured from the respective photometry image.

For robust selection of candidate galaxies at z = 7.7 we followed a set of criteria that has yielded a high spectroscopic success rate at z = 4.5 and 5.7 (Rhoads & Malhotra 2001; Rhoads et al. 2003; Dawson et al. 2004, 2007; Wang et al. 2009). Following this, each of our candidates had to satisfy all of the following criteria: (1) 5σ detection in the NB, (2) 3σ significant narrowband excess (compared to the F125W image), and (3) non-detection ($\lt 2\sigma$) in the individual optical images (F606W, F814W). Criteria 1 and 2 ensure real emission-line sources, while criterion 3 eliminates most low-redshift sources. Using this set of criteria we identified three z = 7.7 candidates, ${\rm{z}}8\_5$, ${\rm{z}}8\_4$, and ${\rm{z}}8\_\mathrm{SM}$, for spectroscopic follow-up. Formally, ${\rm{z}}8\_\mathrm{SM}$ has an NB S/N ∼ 4 (just below our selection threshold); however, it has an aggregate S/N = 13 in NB+F125W+F160W. Given that it satisfied all other selection criteria, and given its proximity to other two galaxies, we included it for the spectroscopic follow-up observations. Galaxy ${\rm{z}}8\_5$ was independently identified as a Lyman-break galaxy by Oesch et al. (2015).

Figure 1 shows image cutouts of all three candidates in five filters. All three galaxies have significant fluxes in the NB filter, indicating the presence of strong emission lines, most likely Lyα lines at the observed wavelength of $1.066\,\mu {\rm{m}}$. None of the galaxies have detectable fluxes at the visible wavelengths, suggesting that these galaxies are consistent with being at redshifts $z\geqslant 7$. All three galaxies are detected in both F125W and F160W (see Table 1), despite two of them being faint, making ${\rm{z}}8\_4$ the faintest galaxy discovered at such high redshifts, thanks to the NB selection technique in which detection of galaxies does not depend on the continuum brightness.

Zoom In Zoom Out Reset image size

Table 1.  Photometry and Spectroscopic Measurements of EGS77

ID	R.A.	Decl.	F606W	F814W	NB	F125W	F160W
	J2000	J2000	(mag)	(mag)	(mag)	(mag)	(mag)
z8_SM	14:20:35.694	+53:00:09.318	<28.3a	<28.2a	24.76 ± 0.35	26.76 ± 0.13	26.66 ± 0.11
z8_4	14:20:35.169	+52:59:40.613	<28.3a	<28.2a	23.85 ± 0.15	27.25 ± 0.21	27.10 ± 0.16
z8_5	14:20:34.872	+53:00:15.242	<28.3a	<28.2a	23.60 ± 0.12	25.29 ± 0.03	25.08 ± 0.03
Spectroscopic Measurements
ID	zspec	${f}_{\mathrm{Ly}\alpha }$	EWrest	${L}_{\mathrm{Ly}\alpha }$	H ii Radiib	S/N	Distancec
		($\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$)	Å	1043($\,\mathrm{erg}\,{{\rm{s}}}^{-1}$)	pMpc	$(\mathrm{Ly}\alpha )$	Trans LoS
z8_SM	7.767	0.29 ± 0.06 × 10−17	23 ± 6	0.2 ± 0.1	0.55	4.9	0.06 0.9
z8_4	7.748	0.56 ± 0.09 × 10−17	71 ± 18	0.4 ± 0.1	0.69	6.0	0.09 0.2
z8_5	7.728	1.70 ± 0.14 × 10−17	37 ± 3	1.2 ± 0.1	1.02	12.1	0.08 −0.5

Notes.

^a ${}^{1}2\sigma$ magnitude limits. All magnitudes are AB magnitudes. ^bH ii bubble radii based on Figure 15 from Yajima et al. (2018). ^cTrans and LoS are the transverse and line-of-sight separation of each galaxy from the flux-weighted mean location of the group center, measured in pMpc.

Download table as:  ASCIITypeset image

2.3. Photometric Redshifts and Spectral Energy Distributions

We use EAZY (Brammer et al. 2008) with spectral energy distribution (SED) templates that have emission lines (eazy_v1.3) along with broadband fluxes to derive the photometric redshift probability distribution p(z). We allow a wide range of redshift grids (z = 0.1 to z = 9) to search for the best-fit SED template. In addition to the photometry (discussed in Section 2.2), we also used HST/WFC3 F105W photometry (GO:13792, PI: Bouwens) for ${\rm{z}}8\_5$ and CFHT-Y band photometry for ${\rm{z}}8\_4$ and ${\rm{z}}8\_\mathrm{SM}$.

As shown in Figure 2, for all three galaxies the best-fit SEDs (shown in blue color) prefer spectral templates that correspond to a photometric redshift ${z}_{\mathrm{phot}}=7.7$. This is also evident from the p(z) (shown in the inset); the presence of strong emission lines in the NB filter yields tight constraints on the p(z). For completeness, we also show the low-redshift galaxy templates (gray color), which are disfavored due to much larger ${\chi }^{2}$ values of the SED fit. For the brightest galaxy ${\rm{z}}8\_5$ (previously identified in Oesch et al. 2015), the difference between the best-fit ${\chi }^{2}$ value of the low-redshift template and high-redshift template is 86. For ${\rm{z}}8\_4$ and ${\rm{z}}8\_\mathrm{SM}$, while the difference between the best-fit ${\chi }^{2}$ value of the low-redshift template and high-redshift template is $\gt 1$, it is not as high as that for ${\rm{z}}8\_5$ given their lower S/N. Furthermore, all three galaxies have large rest-frame Lyα equivalent widths ${\mathrm{EW}}_{\mathrm{rest}}\gt 23$Å (see Table 1), which likely makes them visible as Lyα-emitting galaxies.

Zoom In Zoom Out Reset image size

To confirm the photometric redshifts of these galaxies, we performed Y-band spectroscopy using the Multi-Object Spectrometer for Infra-Red Exploration (MOSFIRE) spectrograph (McLean et al. 2012) on the Keck I telescope. The MOSFIRE instrument allows us to obtain spectroscopic observations of multiple objects simultaneously, with the Y-band covering Lyα lines redshifted to z = 7–8.2. We targeted three Lyα galaxy candidates in the EGS field as our primary science targets, and used low-redshift emission-line candidates as fillers. Observations were taken on 2017 May 6, with individual exposures having 140 sec integration time and AB pattern dither offsets along the slit, with offset of ±1'' from the center. The spectroscopic conditions were good, with a typical seeing of ∼07 and a total integration time of about 4 hr per object.

We reduced data using the public MOSFIRE data reduction pipeline. It performs standard data reduction procedures, including sky subtraction, rectification of the 2D spectra, and wavelength calibration. For a given object, it also produces a corresponding S/N and a sigma image. The final 2D science spectra have a spatial resolution of 018 per pixel and a dispersion of 1.086 Å per pixel. For absolute flux calibration of 1D spectra, we compared the measured line flux of ${\rm{z}}8\_5$ with the calibrated flux from Oesch et al. (2015), and converted instrumental flux (in counts) to the absolute flux. The conversion factor derived for ${\rm{z}}8\_5$ is then used to calibrate spectra of ${\rm{z}}8\_\mathrm{SM}$ and ${\rm{z}}8\_4$.

Figure 3 shows the final coadded 2D and 1D Y-band spectra of all three galaxies. As seen, both galaxies ${\rm{z}}8\_5$ and ${\rm{z}}8\_4$ show prominent emission lines (S/N > 5), while ${\rm{z}}8\_\mathrm{SM}$ has S/N = 4.9. In the following we show that these are very likely Lyα emission at z = 7.7.

Zoom In Zoom Out Reset image size

For both ${\rm{z}}8\_5$ and ${\rm{z}}8\_4$, emission lines are free from the night sky OH lines, as shown in the shaded region (middle row). For ${\rm{z}}8\_\mathrm{SM}$, while its emission line is close to a faint night sky line, given its higher S/N compared to the night sky lines, it is very likely a genuine emission line. Furthermore, the observed Y-position of the emission line in the 2D spectrum of ${\rm{z}}8\_\mathrm{SM}$ matches well with that expected based on the slit position in the mask (Figure 3 top panel), supporting our conclusion that this is a real emission line. We note that the presence of a faint night sky line at the position of the Lyα line may possibly affect the Lyα line flux measurements. However, given the faintness of the night sky line, we expect that it will have a minimal impact on the measurement of the Lyα line flux. In the following we demonstrate that these are very likely Lyα lines at z = 7.7.

One of the characteristics that distinguishes Lyα emission from star-forming galaxies and other emission lines is the line asymmetry (Rhoads et al. 2003) or Skewness (Kashikawa et al. 2006). To quantify the Skewness in the line, we calculated the weighted skewness parameter Sw, and found ${Sw}=22\pm 6$Å for ${\rm{z}}8\_5$. This confirms that ${\rm{z}}8\_5$ is a z = 7.7 galaxy because ${Sw}\gt 3$Å is not seen in low-redshift emission lines of [O ii], [O iii], and Hα (Kashikawa et al. 2006). Indeed, ${\rm{z}}8\_5$ was previously identified as a Lyman-break candidate (labeled as EGS-zs8-1), and spectroscopically confirmed as a z = 7.7 galaxy (Oesch et al. 2015). Furthermore, recent H-band spectroscopic observations of this galaxy show the presence of a [C iii] 1909 doublet (Stark et al. 2017). Thus, given all the evidence, ${\rm{z}}8\_5$ is unequivocally a Lyα-emitting galaxy at z = 7.7. For ${\rm{z}}8\_4$ the weighted skewness parameter for the observed emission line is ${Sw}=17\pm 7$ Å, confirming the line is Lyα at z = 7.7. For ${\rm{z}}8\_\mathrm{SM}$, we cannot reliably measure the asymmetry of the line given its lower S/N in the spectrum. However, based on the best-fit SEDs, both ${\rm{z}}8\_4$ and ${\rm{z}}8\_\mathrm{SM}$ favor high-redshift solutions. Furthermore, if these were faint, low-redshift galaxies, the best-fit low-redshift SEDs imply a clear detection in the F606W and F814W filters. Thus, given all the evidence, ${\rm{z}}8\_5$ and ${\rm{z}}8\_4$ are unequivocally at redshifts z = 7.728 and z = 7.748 respectively, and ${\rm{z}}8\_\mathrm{SM}$ is also very likely at redshift z = 7.767.

It is striking that while ${\rm{z}}8\_4$ and ${\rm{z}}8\_\mathrm{SM}$ are faint (${M}_{{UV}}\gt -20.3$ mag), both show Lyα emission lines. Moreover, despite the low number density of such faint galaxies at $z\gt 7$, all three galaxies are spectroscopically confirmed. This high spectroscopic success rate is likely because (1) our NB technique preselects galaxies with detectably strong line emission, and (2) EGS77 likely formed a large ionized bubble, allowing Lyα photons to escape. We discuss this in more detail in the following section.

The visibility of Lyα emission from star-forming galaxies at high redshifts depends on several factors including star formation rate, ionizing photon budget, galactic outflows, and proximity to ionized bubbles formed by other galaxies (e.g., Matthee et al. 2020). The star formation rate and the ionizing photon budget will directly influence the amount of ionized gas forming an ionized bubble, which in turn allows Lyα photons to travel unattenuated along the line of sight (Malhotra & Rhoads 2006). While there are arguments about whether fainter LBGs (Stark et al. 2011) or brighter LBGs (e.g., Matthee et al. 2020) are more likely to have a higher Lyα escape fraction, our observations (discussed below) show that perhaps the proximity to other galaxies with ionizing flux is the determining factor.

The separation between the most distant member ${\rm{z}}8\_\mathrm{SM}$ (z = 7.767) and ${\rm{z}}8\_4$ (z = 7.748) along the line of sight is about 0.7 pMpc (physical Mpc), which is the same as the separation between ${\rm{z}}8\_4$ and ${\rm{z}}8\_5$ (z = 7.728). In the transverse direction (i.e., projection on the sky) all three members are much closer to each other. The separation between ${\rm{z}}8\_\mathrm{SM}$ and ${\rm{z}}8\_5$ in the transverse direction is 10'' (0.05 pMpc), while the separation between ${\rm{z}}8\_4$ and ${\rm{z}}8\_5$ is 35'' (0.18 pMpc). Thus, given the proximity of all three galaxies with each other in both the transverse direction and along the line of sight, these galaxies will form a continuous attenuation-free path for Lyα photons if the radii of their ionized bubbles are ≥0.35 pMpc along the line-of-sight direction (Figure 4). This is because the ionized region is large enough that the Lyα photons are redshifted by the time they reach the neutral hydrogen boundary, and thus can escape. This is supported by recent spectroscopic observations of the brightest galaxy ${\rm{z}}8\_5$, showing a [C iii] 1909 doublet (Stark et al. 2017), which yields a velocity offset ${\rm{\Delta }}{V}_{\mathrm{Ly}\alpha }$ = 340${}_{-30}^{+15}\,\mathrm{km}\,{{\rm{s}}}^{-1}$. When compared to a FWHM = 360 $\mathrm{km}\,{{\rm{s}}}^{-1}$ for this line (Oesch et al. 2015), it is implied that a substantial fraction of Lyα photons are leaving the galaxy at 340–520 $\mathrm{km}\,{{\rm{s}}}^{-1}$.

Zoom In Zoom Out Reset image size

4.1. Estimation of Bubble Sizes

In this Letter we report the discovery of the farthest galaxy group EGS77 at z = 7.7, merely 680 Myr after the big bang. EGS77 was initially identified using NB imaging observations from the DAWN survey, and later confirmed via spectroscopic observations using the MOSFIRE spectrograph on the Keck telescope. It is striking that all three galaxies in EGS77 are spectroscopically confirmed via Lyα emission line (${\rm{S}}/{{\rm{N}}}_{\mathrm{Ly}\alpha }\gtrsim 5$) despite two of them being faint in the continuum. In fact, EGS77 contains the faintest galaxy (in terms of the continuum brightness) discovered and spectroscopically confirmed at this redshift. Based on models and simulations from Yajima et al. (2018), we found that a large ionized bubble produced by the brightest galaxy ${\rm{z}}8\_5$ and two relatively smaller bubbles surrounding fainter galaxies, overlap significantly, producing a large yet localized ionized region.

At redshift z = 7.7, EGS77 is the most distant spectroscopically confirmed galaxy group yet identified. The associated bubble, implied by the prominent Lyα emission in all three spectra, is the most distant ionized bubble identified, and the first at a redshift where the bulk neutral hydrogen fraction is thought to approach or perhaps even exceed 50% (Robertson et al. 2015; Planck Collaboration et al. 2016; Finkelstein et al. 2019). This combination of local ionization at a redshift with substantially neutral bulk gas provides observational support for the picture of spatially inhomogeneous reionization that proceeds from the growth of ionized bubbles.

We thank the referee for many constructive comments that improved this manuscript. We thank the US NSF (Grant AST-1518057), and NASA for its financial support through the WFIRST Science Investigation Team program (NNG16PJ33C). Z.Y.Z. thanks the NSF of China (11773051) and the CAS Pioneer Hundred Talents Program. Data presented herein were obtained at the W.M. Keck Observatory, operated as a scientific partnership among Caltech, the University of California and the NASA. The Observatory was made possible by the generous financial support of the W.M. Keck Foundation. The authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain.