A Lyman-α protocluster at redshift 6.9
Weida Hu, Junxian Wang, Leopoldo Infante, James E. Rhoads, Zhen-Ya Zheng, Huan Yang, Sangeeta Malhotra, L. Felipe Barrientos, Chunyan Jiang, Jorge González-López, Gonzalo Prieto, Lucia A. Perez, Pascale Hibon, Gaspar Galaz, Alicia Coughlin, Santosh Harish, Xu Kong, Wenyong Kang, Ali Ahmad Khostovan, John Pharo, Francisco Valdes, Isak Wold, Alistair R. Walker, XianZhong Zheng
LLETTER
A Lyman- α PROTOCLUSTER AT REDSHIFT6.9
Weida Hu , , Junxian Wang , , Leopoldo Infante , , , James E. Rhoads , Zhen-Ya Zheng , Huan Yang , Sangeeta Malhotra , L.Felipe Barrientos , Chunyan Jiang , Jorge Gonz´alez-L´opez , , Gonzalo Prieto , Lucia A. Perez , Pascale Hibon , Gaspar Galaz ,Alicia Coughlin , Santosh Harish , Xu Kong , , Wenyong Kang , , Ali Ahmad Khostovan , John Pharo , Francisco Valdes ,Isak Wold , Alistair R. Walker , XianZhong Zheng CAS Key Laboratory for Research in Galaxies and Cosmology, Departmentof Astronomy, University of Science and Technology of China, Hefei, Anhui230026, China; [email protected], [email protected] School of As-tronomy and Space Science, University of Science and Technology of China,Hefei 230026, China Las Campanas Observatory, Carnegie Institution ofWashington, Casilla 601, La Serena, Chile Astrophysics Science Division,Goddard Space Flight Center, 8800 Greenbelt Road, Greenbelt, Maryland20771, USA CAS Key Laboratory for Research in Galaxies and Cosmol-ogy, Shanghai Astronomical Observatory, Shanghai 200030, China Institutode Astrof´ısica, Facultad de F´ısica, Pontificia Universidad Cat´olica de Chile,Santiago, Chile School of Earth and Space Exploration, Arizona State Uni-versity, Tempe, AZ 85287, USA European Southern Observatory, Alonsode Cordova 3107, Casilla 19001, Santiago, Chile N´ucleo de Astronom´ıade la Facultad de Ingenier´ıa y Ciencias, Universidad Diego Portales, Av.Ej´ercito Libertador 441, Santiago, Chile Community Science and Data Cen-ter/NSF’s NOIRLab, 950 N. Cherry Ave., Tucson, AZ 85719, USA CerroTololo Inter-American Observatory, NSF’s NOIRLab, Casilla 603, La Serena,Chile Purple Mountain Observatory, Chinese Academy of Sciences, Nan-jing 210023, China
Protoclusters, the progenitors of the most massive structures inthe Universe, have been identified at redshifts up to 6.6 [refs ].Besides exploring early structure formation, searching for proto-clusters at even higher redshifts is particularly useful to probethe reionization. Here we report the discovery of the protoclus-ter LAGER- z .The protocluster is identified by an overdensity of 6 times the av-erage galaxy density, and with 21 narrowband selected Ly α galax-ies, among which 16 have been spectroscopically confirmed. Atredshifts similar to or above this record, smaller protogroups withfewer members have been reported . LAGER- z . × solar masses. The total volume of the ionized bubblesgenerated by its member galaxies is found to be comparable to thevolume of the protocluster itself, indicating that we are witness-ing the merging of the individual bubbles and that the intergalacticmedium within the protocluster is almost fully ionized. LAGER- z High redshift Lyman- α (Ly α )-emitting galaxies (LAEs) are star-forming galaxies with strong Ly α lines, which can be effectively se-lected with narrowband imaging surveys . Aiming to build a statis-tical sample of LAEs at redshift ∼
7, we are carrying out a deep narrow-band imaging survey, Lyman Alpha Galaxies in the Epoch of Reion-ization (LAGER), utilizing the Dark Energy Camera (DECam, with afield of view of ∼ deg ) on Cerro Tololo Inter-American Observa-tory (CTIO) Blanco 4m Telescope and a customized narrowband filter Wavelength( ˚ A) o f L A E s . . . . T r a n s m i ss i o n DECam - NB964HSC - NB973 z Figure 1 | Redshift distribution of spectroscopically confirmed LAEs inLAGER- z The red histogram shows the redshift distribution of 16 spec-troscopically confirmed LAEs in LAGER- z DECam-NB964. The central wavelength and full-width half-maximumof the filter are ∼ ˚A and 92 ˚A (see Fig. 1), corresponding to a red-shift range of 6.89 – 6.97 and a line-of-sight (LOS) scale of 26 cMpc.See Methods for more details. In the LAGER COSMOS field, we ob-tained 47.25 hours narrowband exposure reaching a 5 σ detection limitof 25.2 magnitude and a Ly α sensitivity of . erg s − . Combin-ing the deep narrowband image with the ultra deep broadband imagesfrom the Hyper Suprime-Cam Subaru Strategic Program (HSC SSP),we uniformly selected 49 z ∼ LAEs . See Methods and papers (hereafter Z17 and H19, respectively) for more details about the LAEselection.As narrowband imaging can constrain the redshift of LAEs to avery narrow range ∆ z < z = 6 – 8, it is also a promising approachto search for overdense structures, for example, protoclusters, in theearly Universe . Fig. 2b shows the spatial distribution (blue circles)and number density (blue contours) of 49 LAGER z ∼ LAEs inthe whole COSMOS field as presented in H19. A high number den-sity region (as marked by black dashed rectangle) is clearly revealed,containing 14 uniformly selected LAEs in H19 (see Suppl. Tab. 1for the catalog). This overdense region (LAGER- z . (cid:48) × (cid:48) , and a three-dimensional volume of × × cMpc . We calculate the galaxy overdensity of LAGER- z δ g = ( n − ¯ n ) / ¯ n , where n and ¯ n are the average number1 a r X i v : . [ a s t r o - ph . GA ] J a n . . . . . R . A . . . . D ec . . ×
21 01 234 5 6 79 1011 13 15171819 812 141620 2223 24 a
10 cMpc 150 . . . . R . A . . . . . D ec . b Figure 2 | Two-dimensional spatial distribution of LAEs in LAGER- z z ∼ . . a, The spatial distribution of member LAEs of LAGER- z b ). Red symbols mark 21 member LAEs of LAGER- z ] in the area which are not detected in NB964 are marked as dashed blue circles. Theyare likely at higher redshifts (beyond the probe of DECam-NB964) and are not considered as members of LAGER- z . (cid:48) × (cid:48) ,corresponding to ∼ ×
30 cMpc ) represents the protocluster region. Note that while all 21 red symbols (open and solid) are considered as members ofLAGER- z (cid:48) × (cid:48) . b, The spatial distribution of 49 LAGER z ∼ densities of LAEs in the LAGER- z z δ g = 5 . +2 . − . , which indicates LAGER- z , the bandpass of which partially overlaps with that ofDECam-NB964 (see Fig. 1). A hint of overdensity around LAGER- z , butnot as strong as that seen in DECam-NB964. We stacked the DECam-NB964 image and HSC-NB973 image to improve the depth of the nar-rowband image and selected three more members of LAGER- z z .We carried out new spectroscopic followups using the Inamori Magel-lan Areal Camera and Spectrograph (IMACS) at the 6.5m Magellan IBaade Telescope (Feb. 6-8, 2017 and Feb. 21-23, 2018), and the LowDispersion Survey Spectrograph 3 (LDSS3) at the 6.5m Magellan IIClay Telescope (January 10-11 and December 29-31, 2019). The aver-age seeing during the observations was ∼ . (cid:48)(cid:48) . We carefully reducedthe observed data and ruled out foreground identifications for membergalaxies. Details of the data reduction are presented in Methods and a dedicated spectroscopic paper in preparation (along with identifica-tions of LAEs outside of LAGER- z α lines were not detected in threeadditional members which were put on masks, however we are unableto rule them out as their Ly α lines might incidentally overlap with skylines, or their Ly α line width be too broad to be detected ( >
500 kms − ). The two- and one-dimensional spectra of the confirmed LAEsare presented in the Suppl. Fig. 1 and the redshift distribution in Fig.1. The scale ( × × cMpc ), overdensity ( δ g = 5 . +2 . − . ),and LOS velocity dispersion of spectroscopic confirmations ( ∼ km s − ) of our protocluster LAGER- z ] andsimulation predictions . We estimate the total present-day clustermass M z =0 of LAGER- z : M z =0 = (1 + δ m )¯ ρV , where V is the volume of the protocluster, ¯ ρ ( . × M (cid:12) cMpc − ) is the mean matter density of the uni-verse, and δ m the mass overdensity. δ m is related to the observedgalaxy overdensity through: bδ m = C (1 + δ g ) , where b is thebias parameter and C the correction factor for the redshift space distor-tion. For δ g = 5 . at z ∼ , we find C = 0 . and δ m = 0 . .The present-day mass M z =0 of LAGER- z . +0 . − . × M (cid:12) , comparable to the mass of nearby COMAcluster ( ∼ × M (cid:12) ). See Methods for details.The 3D distribution of the spectroscopic confirmations is shownin Fig. 3. LAGER- z . +3 . − . (left) and . +4 . − . (right), respectively, where the bound-aries of the two substructures are defined as the blue squares in Fig 2.If we treat the two substructures as isolated, their present-day massesare expected to be . +0 . − . × M (cid:12) and . +0 . − . × M (cid:12) ,respectively.We further explore whether the protocluster LAGER- z ec . ( c M p c ) − R . A . ( c M p c ) L O S D i s t a n ce ( c M p c ) − Figure 3 |
3D spatial distribution of 16 spectroscopically confirmed LAEs inLAGER- z z ∼ . . The red points represent the 16 spectroscopicallyconfirmed LAEs in LAGER- z would collapse into a single cluster. Similar to previous works , weestimate the linear overdensity of LAGER- z δ L = 0 . at z ∼ (Equation 18 of ref. ). As the growth of linear perturbation δ L is proportional to t / , δ L will be larger than the threshold δ L > . at z = 2 , where δ L = 1 . is the critical value of linear overdensityof a spherical perturbation at the time it collapses . Thus, we expectLAGER- z z , which can reduce resonant scattering of Ly α photons in the neutral IGM . The protoclusters in the EoR may leadthe production of such bubbles because of their high number densityof galaxies. On the basis of the relation between the bubble size andLy α luminosity in a simulation of reionization work , we show thepredicted bubbles in Fig. 3 and the bubble sizes in Suppl. Tab. 1. Thesummed volume of the ionized bubbles of all 21 LAEs is . × cMpc , with the 4 most luminous ones (with L Lyα > × ergs − , i.e., LAE 1,2,3,15) contributing . of the total ionized vol-ume. This total ionized volume is even slightly larger than the volumeof LAGER- z . × cMpc ). This demonstrates significantoverlaps between individual bubbles, indicating the individual bubblesare in the act of merging into one or two giant bubbles (see Fig. 3).As a comparison, the total predicted volume of all the 49 uniformly se-lected LAEs in COSMOS field is . × cMpc , correspondingto . of the total volume surveyed by DECam-NB964. See Meth-ods for details. Such predicted giant bubbles are large enough to beresolved by future 21-cm programs, e.g., SKA1-Low with resolutionof ∼ . arcmin at z ∼ . [ref. ], corresponding to ∼ cMpc.The merged bubble (with predicted size of (cid:38) cMpc) could sig-nificantly increase the IGM transmission, and thus enhance the Ly α visibility of member LAEs . Note Z17 and H19 have revealed abright-end excess in the Ly α luminosity function in COSMOS field,also suggesting the existence of big ionized bubbles at z ∼ that re-duce the opacity of neutral IGM around the luminous LAEs. Mean-while, if the Ly α transmission through the IGM has been significantly boosted in most LAEs in LAGER- z α equivalent widths (EWs) of LAEs in the protocluster. However, the ex-pected larger Ly α EWs is not seen, compared with the field LAEs inCOSMOS (see Methods and Extended Data Fig. 1 for details), thoughthe large uncertainties in the EW measurements and the small sam-ple size prevent us from reaching a robust conclusion. One possibilityis that high-redshift protoclusters are highly biased regions and mightcontain LAEs with physical properties deviating substantially from thefield LAEs . It is yet unclear if the intrinsic Ly α escape (prior toIGM scattering) in clustered LAEs is the same as that in field LAEs.Moreover, the expected excess of close companions due to poten-tially enhanced Ly α transmission, in or behind the large bubbles of theluminous LAEs ( L Lyα > × erg s − , i.e., LAE 1,2,3,15), isnot seen (see Fig. 2a and 3). This is likely in part due to the possi-bility that while the biased dark matter halos can increase the galaxymerger/interaction, and thus enhance the star formation in the over-dense region , the feedback from the UV background may suppressthe star formation in the nearby fainter galaxies .The discovery of the protocluster LAGER- z α fainter galaxies par-tially responsible for the ionization budget, better constraining the Ly α line EWs and the Ly α profiles, measuring the Ly α velocity offsets rela-tive to their system redshifts, and mapping the star formations historiesof the galaxies. Correspondence and request for materials should be addressed to W.Hu and J. Wang (e-mail: [email protected], [email protected]).
Acknowledgements
We appreciate the anonymous referees for the valu-able comments and Zhen-Yi Cai, Zheng Cai, and Linhua Jiang for the infor-mative discussions. The work is supported by National Science Foundationof China (grants No. 11421303 & & & Facilities:
Magellan:Baade (IMACS), Magellan:Clay (LDSS3), Blanco (DE-Cam), Subaru (HSC)
Author contributions
W.H. and J.W designed the layout of this paper. W.H.reduced the data, performed scientific analysis and wrote the manuscript.J.W. co-led the scientific interpretation and manuscript writing. L.I. led the ob-serving proposals which yielded new spectroscopic identifications presentedin this work. W.H., L.I., H.Y., J. G. & G.P. conducted these observations. Allauthors discussed the results and commented on the manuscript.
Competing Interests
The authors declare that they have no competing fi-nancial interests. ata availability The candidate selection is based on the following im-ages in COSMOS field: DECam-NB964 (NOIRLab Prop. ID: 2016A-0386, 2017B-0330; CNTAC Prop. ID: 2016A-0610), HSC SSP program,and HSC-NB973 (Prop. ID: S16B-001I), which are available at http://archive1.dm.noao.edu/ , https://hsc-release.mtk.nao.ac.jp/doc/ , and https://hsc-release.mtk.nao.ac.jp/doc/index.php/chorus/ , respectively. The spectroscopic datasets and the datasets gener-ated or analysed during this study are available from the corresponding authorupon reasonable request. The LAE catalog for LAGER- z Code Availability Statement
Codes used in this study are not publicly re-leased yet but are available from the corresponding author on reasonablerequest. 1. Ouchi, M. et al.
The Discovery of Primeval Large-Scale Structures withForming Clusters at Redshift 6.
Astrophys. J. , L1–L4 (2005).2. Wang, J. X., Malhotra, S. & Rhoads, J. E. An Overdensity of Ly α Emittersat Redshift z ∼ Astrophys. J. ,L77–L80 (2005).3. Malhotra, S. et al.
An overdensity of galaxies at z= 5.9 ± Astrophys. J. , 666–679 (2005).4. Jiang, L. et al.
A giant protocluster of galaxies at redshift 5.7.
Nat. Astron. , 962–966 (2018).5. Harikane, Y. et al. SILVERRUSH. VIII. Spectroscopic Identifications ofEarly Large-scale Structures with Protoclusters over 200 Mpc at z ∼ Astrophys. J. , 142 (2019).6. Calvi, R. et al.
MOS spectroscopy of protocluster candidate galaxies at z= 6.5.
Mon. Not. R. Astron. Soc. , 3294–3306 (2019).7. Robertson, B. E., Ellis, R. S., Furlanetto, S. R. & Dunlop, J. S. CosmicReionization and Early Star-forming Galaxies: A Joint Analysis of NewConstraints from Planck and the Hubble Space Telescope.
Astrophys. J. , L19 (2015).8. Kulkarni, G. et al.
Large Ly α opacity fluctuations and low CMB τ in mod-els of late reionization with large islands of neutral hydrogen extending toz < Mon. Not. R. Astron. Soc. , L24–L28 (2019).9. Castellano, M. et al.
Spectroscopic Investigation of a Reionized GalaxyOverdensity at z = 7.
Astrophys. J. , L3 (2018).10. Tilvi, V. et al.
Onset of Cosmic Reionization: Evidence of an IonizedBubble Merely 680 Myr after the Big Bang.
Astrophys. J. , L10 (2020).11. Zheng, Z.-Y. et al.
First Results from the Lyman Alpha Galaxies in theEpoch of Reionization (LAGER) Survey: Cosmological Reionization at z ∼ Astrophys. J. , L22 (2017).12. Itoh, R. et al.
CHORUS. II. Subaru/HSC Determination of the Ly α Lu-minosity Function at z = 7.0: Constraints on Cosmic Reionization ModelParameter.
Astrophys. J. , 46 (2018).13. Hu, W. et al.
The Ly α Luminosity Function and Cosmic Reionization atz ∼ Astrophys. J. , 90 (2019).14. Hu, W. et al.
First Spectroscopic Confirmations of z ∼ α EmittingGalaxies in the LAGER Survey.
Astrophys. J. , L16 (2017).15. Overzier, R. A. et al. Λ CDM predictions for galaxy protoclusters - I. Therelation between galaxies, protoclusters and quasars at z ∼ Mon. Not.R. Astron. Soc. , 577–594 (2009).16. Chiang, Y.-K., Overzier, R. & Gebhardt, K. Ancient Light from YoungCosmic Cities: Physical and Observational Signatures of Galaxy Proto-clusters.
Astrophys. J. , 127 (2013).17. Merritt, D. The Distribution of Dark Matter in the Coma Cluster.
Astro-phys. J. , 121 (1987).18. Cai, Z. et al.
Mapping the Most Massive Overdensities through Hydro-gen (MAMMOTH). II. Discovery of the Extremely Massive OverdensityBOSS1441 at z = 2.32.
Astrophys. J. , 131 (2017).19. Chanchaiworawit, K. et al.
Physical Properties of a Coma-analog Proto-cluster at z = 6.5.
Astrophys. J. , 51 (2019).20. Mo, H. J. & White, S. D. M. An analytic model for the spatial clustering ofdark matter haloes.
Mon. Not. R. Astron. Soc. , 347–361 (1996).21. Jenkins, A. et al.
The mass function of dark matter haloes.
Mon. Not. R.Astron. Soc. , 372–384 (2001).22. Gnedin, N. Y. Cosmological Reionization by Stellar Sources.
Astrophys.J. , 530–554 (2000).23. Rodr´ıguez Espinosa, J. M. et al.
An ionized superbubble powered by aprotocluster at z = 6.5.
Mon. Not. R. Astron. Soc. , L17–L21 (2020).24. Malhotra, S. & Rhoads, J. E. The Volume Fraction of Ionized IntergalacticGas at Redshift z=6.5.
Astrophys. J. , L95–L98 (2006).25. Dijkstra, M., Lidz, A. & Wyithe, J. S. B. The impact of The IGM on high-redshift Ly α emission lines. Mon. Not. R. Astron. Soc. , 1175–1186(2007).26. Yajima, H., Sugimura, K. & Hasegawa, K. Modelling of Lyman-alphaemitting galaxies and ionized bubbles at the epoch of reionization.
Mon.Not. R. Astron. Soc. , 5406–5421 (2018).27. Wyithe, S., Geil, P. & Kim, H. Imaging HII Regions from Galaxiesand Quasars During Reionisation with SKA.
Proc. Sci.
AASKA14 , 015(2015).28. Yajima, H., Shlosman, I., Romano-D´ıaz, E. & Nagamine, K. Observa-tional properties of simulated galaxies in overdense and average regionsat redshifts z (cid:39)
Mon. Not. R. Astron. Soc. , 418–432 (2015).29. Lee, C. T. et al.
Properties of dark matter haloes as a function of localenvironment density.
Mon. Not. R. Astron. Soc. , 3834–3858 (2017).30. Maio, U., Petkova, M., De Lucia, G. & Borgani, S. Radiative feedbackand cosmic molecular gas: the role of different radiative sources.
Mon.Not. R. Astron. Soc. , 3733–3752 (2016). ETHODS
Throughout this study, we adopt the recent Planck cosmologicalparameters : Ω m = 0 . , Ω Λ = 0 . and H = 67 . km s − Mpc − , where Ω m and Ω Λ are the densities of total matter and dark energyand H is the Hubble constant. Lyman Alpha Galaxies in the Epoch of Reionization (LAGER)survey:
LAEs are promising probes for characterizing the cosmicreionization . We are carrying out a large-area narrowband imag-ing survey, Lyman Alpha Galaxies in the Epoch of Reionization (LAGER),to search for the LAEs at z ∼ . , using the Dark Energy Camera (DECam)installed on the Cerro Tololo Inter-American Observatory (CTIO). Wedesigned and procured a narrowband filter (DECam-NB964) for the LAGERsurvey, with a central wavelength of ∼ ˚A and FWHM of ∼
92 ˚A toavoid the atmospheric absorption and strong OH emission lines. The filterDECam-NB964 was installed on the DECam system in December 2015.Owing to the large FoV ( ∼ deg ) and red-sensitive camera, LAGER isone of the most efficient surveys in searching for LAEs in the EoR. See thefilter design paper for more details. We adopt a ”wedding cake” observingstrategy with two deep fields aiming to discover faint LAEs and severalshallower fields aiming to discover numerous luminous LAEs. Up to now,candidate LAEs have been selected in 4 fields, including COSMOS, CDFS,WIDE12, and GAMA15A. LAGER- z The 14 LAEs from H19 were selected by narrowband technique. Thistechnique is widely used in literature and has been proven effective atsearching for LAEs. Briefly, the selection criteria in H19 include: (1) thesignal-to-noise ratio (S/N) of DECam-NB964 signal is larger than 5; (2)DECam-NB964 excess over the underlying broadband to ensure the restframe equivalent width (EW) of Ly α is larger than 10 ˚A; (3) non-detectionin bluer broadbands (we adopt the recently release HSC SSP ultradeepbroadband images ). We visually inspected each LAE candidate to removepossible foreground galaxies and spurious objects, such as satellite trails,cosmic rays, etc. Finally, we obtained a clean sample of 49 LAEs inCOSMOS field and a clear overdense region is revealed with 14 LAEs. b. Additional members: The same field was also observed with another narrowband filterHSC-NB973 , the bandpass of which partially overlaps with that ofDECam-NB964 (see Fig. 1). A hint of over-density around LAGER- z , but not as strongas that seen in DECam-NB964. Among the 8 HSC-NB973 selected LAEsin the area, four (LAE-1,2,11,15) were detected in DECam-NB964 andpresented by H19. An additional source (LAE-20) shows tentative signal( S/N = 4 . ) in DECam-NB964 image (see also next paragraph), whilethe remaining 3 (LAE-22,23,24; blue dashed circles in Fig. 2; namelyHSC-z7LAE24,6,16 respectively in ref. ) are invisible in DECam-NB964.The latter three had not been spectroscopically observed, and are candidateLAEs likely at slightly higher redshifts beyond the probe of NB964 (seeFig. 1). At the current stage, we do not consider these 3 LAEs as membergalaxies of LAGER- z The overdensity in LAGER- z We estimate the overdensity as de-fined by δ g = ( n − ¯ n ) / ¯ n , where n and ¯ n are the average LAE numberdensities in the LAGER- z ¯ n ∼ . +0 . − . arcmin − (49 over 1.9 deg ) and the number density of LAEs in the LAGER- z n ∼ . +0 . − . arcmin − (14 over 0.088 deg ). The errors arecalculated based on the Poisson errors of the LAE sample size. The galaxyoverdensity of LAGER- z δ g = 5 . +2 . − . . The LAE samplesuffers incompleteness during the detection and selection procedures (seeH19 for details). However, as the narrow- and broad-band images utilizedfor LAE detection and selection have rather uniform depths throughout theCOSMOS field, the incompleteness is constant over the field, and thus, can-cels out in the calculation of the overdensity. Note we use only the LAEsample in COSMOS field from H19 for the calculation of overdensity. Theadditional members to LAGER- z z ∼ of the volume. If we exclude this void region, the overdensity would beincreased to ∼ σ Poisson errors (Woldet al. in preparation). We integrate the LFs in the luminosity range of . − . erg s − and find the derived average LAE densitiesfrom the four fields agree within . Thus the field-to-field variation hasno significant effect on the calculation of overdensity.We finally note that we can not rule out the possibility that a smallfraction of the uniformly selected 49 candidate from H19 are actually notreal LAEs, but contaminants (such as noise spikes in narrowband image,variable sources, or foreground emission line galaxies). The total numberof contaminated foreground emission line galaxies (H α , [OIII] and [OII])was estimated to be 0.82 in COSMOS, thanks to the ultradeep broadbandimages available . Considering the contaminants are unlikely spatiallyassociated with the protocluster and should distribute randomly, excludingsuch contaminants (even if possible) from the calculation would yield evenhigher overdensity. Spectroscopic observation and data reduction:
The three brightest LAEsin LAGER- z ]. We carried out spectroscopic followups for moreLAEs in LAGER- z (cid:48) diameter) and the 300-linered-blazed grism. For LDSS3, we used VPH-Red grism and OG590 filterto eliminate second order contamination. Comparing with IMACS, LDSS3has a smaller FoV ( . (cid:48) diameter) but relatively higher efficiency at 9600– 9700 ˚A. Slitwidth of (cid:48)(cid:48) was adopted for both instruments. The spectralreduction was performed using COSMOS3 and the single-epoch spec-tra are average with weights selected to maximize the
S/N of the coaddedspectra.The result 1D spectra (both IMACS and LDSS3) have spectral resolu-tion of ∼ , we now have spectroscopic con-firmations for 16 member LAEs in LAGER- z aminated by foreground strong emission line galaxies, e.g., H α , [O III ], and[O II ]. Due to the limited spectral quality (and partial overlap with sky linesfor some of them) we are unable to secure the characteristic asymmetric lineprofile (with a red wing) of high-z Ly α lines for many sources. Mean-while while some lines are too narrow to be [O II ] doublet, for some broaderones [O II ] can not be completely ruled out based on the line profile alone .However, the contamination rate is expected to be low. For example, recentspectroscopic survey of high-redshift LAEs at z ∼ . reports a low con-tamination rate of < in their spetroscopic detections . Even we con-sider a contamination rate of 10%, our single line identifications would bereliable for most sources. More critically and fortunately, in COSMOS ultradeep broadband images are available to rule out almost all low-z interlopersof emission line galaxies, and we expect the sample of H19 include only ∼ II ]), 0.52 ([O III ]), and 0.16 (H α ) low-z emission line galaxies overthe whole COSMOS field .LAE-8, 12, and 14 were also spectroscopically observed, but notyet confirmed. The non-detections of the Ly α in their spectra do notnecessarily rule them out, as their Ly α lines might incidentally overlapwith sky lines, or the velocity dispersions of their Ly α lines could be toobroad to be detected ( >
500 km s − ). Note we do not detect either anysignals (lines, continuum) indicative of foreground sources in their spectra.These candidates which are spatially associated with the protocluster aremore likely real LAEs instead of contaminations (such as variable sourcesor noise spikes in the narrowband image). This is because the area ofLAGER- z ∼ z Present-day mass of LAGER- z We estimate the total present-day cluster mass M z =0 of LAGER- z : M z =0 = (1 + δ m )¯ ρV , where V is the volume of theprotocluster, ¯ ρ (3 . × M (cid:12) ) is the mean matter density of theuniverse, and δ m is the mass overdensity. δ m is related to the observedgalaxy overdensity through: bδ m = C (1 + δ g ) , where b is thebias parameter and C the correction factor for the redshift space dis-tortion, C = 1 + f − f (1 + δ m ) / and f = Ω m z / . The biasparameter is assumed to be b = 4 . ± . (measured using z ∼ . LAEs ). For δ g = 5 . at z ∼ , we find C = 0 . and δ m = 0 . .Thus, the present-day mass M z =0 of LAGER- z . +0 . − . × M (cid:12) (where the errors are derived through simulatingthe fluctuations of the galaxy overdensity δ g and bias parameter b ). Asaforementioned, the selection of the boundary is kind of arbitrary andcould introduce as much as uncertainty to the overdensity estimation.However, this effect is moderate when estimating the present-day mass ofthe LAGER- z would yield a lower M z =0 and increasing the volume by would yield a higher M z =0 . The bias parameter might have been underestimated since weadopted the value at z ∼ . , and specifically, an increase of b from 4.5 to5.5 will result in a ∼ decrease of the estimated M z =0 . Bubble size estimation:
Previous studies presented a semi-numericalsimulation to investigate the relation between the bubble size and Ly α lu-minosity of high-redshift LAEs in EoR. In the simulation, the star formationrate (SFR) was assumed proportional to the growth rate of the dark matterhalo, the escape fraction of ionizing photons was assumed to be 0.2, andthe ionizing photon emissivity was calculated based on the star formationhistory of the galaxy using the population synthesis code STARBURST99 .Finally, the evolution of ionized bubble and Ly α luminosity (derived fromthe ionizing photons which do not escape) were obtained after calculatingthe radiative transfer in the IGM. Based on the relation between the bubblesize and Ly α luminosity (at z = 8 , Fig. 15 in ref. ), we show the predicted . . . . . HSC - y − NB964 (mag) . . . . . . F r a c t i o n LAGER - z
10 20 40 80 160 640
EW ( ˚ A) Extended Data Figure 1 | The HSC- y – DECam-NB964 (and Ly α EW) distri-bution of LAEs in the LAGER COSMOS field.
The LAEs inside the LAGER- z y we simply adopt the 2 σ lower limits to their HSC- y magni-tudes. Most sources with color > y . Thevertical lines plot the median colors (1.84 and 1.86) respectively. The tick mark ofLy α EW is derived from the color assuming a redshift of 6.931 (corresponding tothe center of NB964 transmission curve). bubble (as translucent spheres) in Fig. 3 and the predicted bubble size inSuppl. Tab. 1 for the spectroscopically confirmed LAEs in LAGER- z has assumeda constant mean IGM density outside an HII bubble with a clumping factor of C = 3 considered to take account of the density fluctuation. While increasingthe clumpiness C would not significantly decrease the bubble size , ref. has pointed out that the higher IGM density near the virial radius may reducethe predicted bubble sizes. The predicted bubble size is sensitive to the Ly-man continuum escape fraction which was assumed to be a constant of 0.2 inref. . But note the observational results at z < suggest that only a smallfraction of galaxies has a high escape fraction of > , and the ioniz-ing continuum escape fraction could be mass-dependent . Moreover, inthe model of ref. , the Ly α escape fraction was assumed to be a constantof 0.6, and both the bubble size and Ly α luminosity tightly correlate withgalaxy stellar mass. However, it is known that high redshift LAEs on aver-age are low mass and young galaxies, i.e., the Ly α escape fraction is massand stellar age dependent (e.g. refs ). Consequently, our LAEs could besignificantly less massive and younger than the model predictions of ref. ,and thus would be expected to have considerably smaller bubble sizes.Nevertheless, considering the mean neutral hydrogen fraction at z ∼ . ( x HI = z z z ∼ . . However, it is yet uncertain whetherthe member LAEs we detected alone can produce such a giant bubble, astheir predicted bubble sizes are remarkably model dependent. In the casesaforementioned, more undetected Ly α fainter galaxies (with lower starformation rates and/or Ly α escape fraction) could have contributed to thereionization around LAGER- z EW/color distribution:
The Ly α EWs of narrowband selected LAEs canbe well represented by the color between narrowband and the underlyingbroadband. For LAEs with spectroscopic redshifts one could derive moreprecise Ly α EW measurements, after correcting for the wavelength depen-dence (non-boxcar shape) of the narrowband transmission and the redshiftdependence of continuum contribution to narrowband photometry (see Fig.2 of H19). As not all LAEs (particularly those field LAEs) have spectro-scopic redshifts, here, we simply use the HSC- y – DECam-NB964 color asan indicator of Ly α EW and compare the color distribution of LAEs insidethe LAGER- z
1. Planck Collaboration et al.
Planck 2018 results. VI. Cosmological param-eters.
Astron. Astrophys. , A6 (2020).32. Malhotra, S. & Rhoads, J. E. Luminosity Functions of Ly α Emitters atRedshifts z=6.5 and z=5.7: Evidence against Reionization at z ≤ As-trophys. J. , L5–L8 (2004).33. Furlanetto, S. R., Zaldarriaga, M. & Hernquist, L. The effects of reioniza-tion on Ly α galaxy surveys. Mon. Not. R. Astron. Soc. , 1012–1020(2006).34. McQuinn, M., Hernquist, L., Zaldarriaga, M. & Dutta, S. Studying reion-ization with Ly α emitters. Mon. Not. R. Astron. Soc. , 75–96 (2007).35. Ouchi, M. et al.
Statistics of 207 Ly α Emitters at a Redshift Near 7:Constraints on Reionization and Galaxy Formation Models.
Astrophys.J. , 869–894 (2010).36. Hu, E. M. et al.
An Atlas of z = 5.7 and z = 6.5 Ly α Emitters.
Astrophys.J. , 394–423 (2010).37. Jiang, L. et al.
A Magellan M2FS Spectroscopic Survey of Galaxies at5.5 < z < α Emitters.
Astrophys. J. , 134 (2017).38. Chanchaiworawit, K. et al.
Gran Telescopio Canarias observations of anoverdense region of Lyman α emitters at z = 6.5. Mon. Not. R. Astron.Soc. , 2646–2661 (2017).39. Konno, A. et al.
SILVERRUSH. IV. Ly α luminosity functions at z = 5.7and 6.6 studied with ∼ α emitters on the 14-21 deg sky. Publ.Astron. Soc. Jpn. , S16 (2018).40. Higuchi, R. et al. SILVERRUSH. VII. Subaru/HSC Identifications of Pro-tocluster Candidates at z ∼ Astrophys. J. , 28 (2019).41. Taylor, A. J., Barger, A. J., Cowie, L. L., Hu, E. M. & Songaila, A. TheUltraluminous Ly α Luminosity Function at z = 6.6.
Astrophys. J. , 132(2020).42. Jung, I. et al.
Texas Spectroscopic Search for Ly α Emission at the End ofReionization III. the Ly α Equivalent-width Distribution and Ionized Struc-tures at z > . arXiv e-prints arXiv:2009.10092 (2020).43. Zheng, Z.-Y. et al. Design for the First Narrowband Filter for the DarkEnergy Camera: Optimizing the LAGER Survey for z ∼ Publ.Astron. Soc. Pacific , 074502 (2019).44. Aihara, H. et al.
First data release of the Hyper Suprime-Cam SubaruStrategic Program.
Publ. Astron. Soc. Jpn. , S8 (2018).45. Dressler, A. et al. IMACS: The Inamori-Magellan Areal Camera andSpectrograph on Magellan-Baade.
Publ. Astron. Soc. Pacific , 288(2011).46. Oemler, A., Clardy, K., Kelson, D., Walth, G. & Villanueva, E. COSMOS:Carnegie Observatories System for MultiObject Spectroscopy (2017).47. Kashikawa, N. et al.
The End of the Reionization Epoch Probed by Ly α Emitters at z = 6.5 in the Subaru Deep Field.
Astrophys. J. , 7–22(2006).48. Ning, Y. et al.
The Magellan M2FS Spectroscopic Survey of High-redshiftGalaxies: A Sample of 260 Ly α Emitters at Redshift z ≈ Astrophys.J. , 4 (2020).49. Steidel, C. C. et al.
A Large Structure of Galaxies at Redshift Z approxi-mately 3 and Its Cosmological Implications.
Astrophys. J. , 428–438(1998).50. Ouchi, M. et al.
Systematic Identification of LAEs for Visible Explorationand Reionization Research Using Subaru HSC (SILVERRUSH). I. Pro-gram strategy and clustering properties of ∼ α emitters at z =6–7 over the 0.3–0.5 Gpc survey area. Publ. Astron. Soc. Jpn. , S13(2018).51. Leitherer, C. et al. Starburst99: Synthesis Models for Galaxies with ActiveStar Formation.
Astrophys. J. Suppl. Ser. , 3–40 (1999).52. Cen, R. & Haiman, Z. Quasar Str¨omgren Spheres Before CosmologicalReionization.
Astrophys. J. , L75–L78 (2000).53. Vanzella, E. et al.
The Great Observatories Origins Deep Survey: Con-straints on the Lyman Continuum Escape Fraction Distribution of Lyman-break Galaxies at 3.4 < z < Astrophys. J. , 1011–1031 (2010).54. Grazian, A. et al.
Lyman continuum escape fraction of faint galaxies at z ∼ Astron. Astrophys. , A18 (2017).55. Bian, F. & Fan, X. Lyman continuum escape fraction in Ly α emitters at z (cid:39) Mon. Not. R. Astron. Soc. , L65–L69 (2020).56. Finkelstein, S. L. et al.
Conditions for Reionizing the Universe with a LowGalaxy Ionizing Photon Escape Fraction.
Astrophys. J. , 36 (2019).57. Ma, X. et al.
No missing photons for reionization: moderate ionizing pho-ton escape fractions from the FIRE-2 simulations.
Mon. Not. R. Astron.Soc. (2020).58. Finkelstein, S. L., Rhoads, J. E., Malhotra, S. & Grogin, N. Lyman AlphaGalaxies: Primitive, Dusty, or Evolved?
Astrophys. J. , 465–481(2009). 59. Lai, K. et al.
Spitzer Constraints on the Stellar Populations of Ly α -Emitting Galaxies at z = 3.1. Astrophys. J. , 70–74 (2008).60. Cai, Z.-Y. et al.
A Physical Model for the Evolving Ultraviolet LuminosityFunction of High Redshift Galaxies and their Contribution to the CosmicReionization.
Astrophys. J. , 65 (2014).
Supplementary Information upplementary Table 1 | Properties of LAEs in the LAGER- z We list the 21 member LAEs in LAGER- z α photometric luminosity. Column 5 lists the redshifts inferred from the line center for spectroscopicconfirmations. Columns 6 – 8 show their AUTO magnitude in the narrowbands DECam-NB964, HSC-NB973, and underlying broadband HSC-y ( σ upper limits fornon-detections). Column 9 lists the bubble size inferred using the relation in ref. . Column 10 is the source IDs in ref. (hereafter H19). ID RA DEC log L Ly α (erg s − ) Redshift DECam-NB964(mag) HSC-NB973(mag) HSC- y (mag) R (cMpc) ID in H19Spectroscopically ConfirmedLAE-1 10:02:06.0 +02:06:46.3 . +0 . − . . ± . c . ± . . +0 . − . . ± . c . ± . . +0 . − . . ± . ** . ± . . +0 . − . . ± . ** . ± . . +0 . − . . ± . ** . ± . a LAE-6 10:03:28.0 +02:08:51.3 . +0 . − . . ± . ** . ± . a LAE-7 10:03:05.2 +02:09:14.7 . +0 . − . . ± .
19 24 . ± .
18 25 . ± . b LAE-9 10:03:16.0 +02:15:42.3 . +0 . − . . ± . ** . ± . a LAE-10 10:02:42.3 +02:06:55.2 . +0 . − . . ± . ** . ± . a LAE-11 10:02:39.4 +02:07:12.1 . +0 . − . . ± . c . ± . . +0 . − . . ± . ** . ± . . d . ± . c . ± . . +0 . − . . ± . ** > . . +0 . − . . ± . ** > . . +0 . − . . ± . ** > . . +0 . − . . ± . ** . ± . . +0 . − . ** . ± . ** . ± . . +0 . − . ** . ± . ** . ± . . +0 . − . ** . ± . ** . ± . . d ** . ± . c . ± . b LAE-21 10:03:15.6 +02:18:11.3 . +0 . − . ** . ± .
20 25 . ± . > . ba LAE-5, 6, 9, 10 were not included in H19 as they are labelled as lower-grade candidates for various reasons (LAE-5: close to bad imageregions; LAE-6: noisy signal in the NB image; LAE-9: adjacent to a foreground galaxy within (cid:48)(cid:48) ; LAE-10: with DECam-NB964 signallower than 5 σ ), but latterly got spectroscopically confirmed. b LAE-7, 20, and 21 are selected using the stacked image of DECam-NB964 and HSC-NB973 images. c We adopt the HSC-NB973 magnitudes given in ref. . d We adopt the Ly α luminosities given in ref , because their Ly α lines locate in the red tail of the DECam-NB964 and their Ly α luminositieswill be severely underestimated if using DECam-NB964 magnitude.8 l u x ( A r b i tr a r y ) LAE − LAE − LAE − LAE − F l u x ( A r b i tr a r y ) LAE − LAE − LAE − LAE − F l u x ( A r b i tr a r y ) LAE − LAE − LAE − LAE − Wavelength ( ˚ A) F l u x ( A r b i tr a r y ) LAE − Wavelength ( ˚ A) LAE − Wavelength ( ˚ A) LAE − Wavelength ( ˚ A) LAE − Supplementary Figure 1 | Two- and one-dimensional spectra of 16 confirmed LAEs in LAGER- z In the top panel the two-dimensional spectra (yellow ishigh flux and blue is low flux) are smoothed by a Gaussian kernel with 1 pixel for better illustration. The two black dashed lines (separated by (cid:48)(cid:48) vertically) representthe expected slit position of LAEs in the 2D spectra. In the middle panel, the blue lines are the one-dimensional spectra and the orange lines are the noise spectra.The grey regions represent the sky OH lines (imperfect sky line subtraction could yield artificial signals visible in the spectra). The dashed horizontal lines indicatezero-flux level and the black arrows mark the peak of the identified Ly α line profiles. In the bottom panel, we plot the S/N spectra with the dashed horizontal linesshowing zero S/N. Due to the flaws in the slit, LAE-16 shows a noisy 2D spectrum. However, a clear broad red wing of the line is revealed which falls in the skylinefree region, and the line is considerably broader than artificial line signals in the spectrum. Thus, we identify it as a Ly α line.line.