Exploring Reionization-Era Quasars III: Discovery of 16 Quasars at 6.4≲z≲6.9 with DESI Legacy Imaging Surveys and UKIRT Hemisphere Survey and Quasar Luminosity Function at z∼6.7
Feige Wang, Jinyi Yang, Xiaohui Fan, Xue-Bing Wu, Minghao Yue, Jiang-Tao Li, Fuyan Bian, Linhua Jiang, Eduardo Bañados, Jan-Torge Schindler, Joseph R. Findlay, Frederick B. Davies, Roberto Decarli, Emanuele P. Farina, Richard Green, Joseph F. Hennawi, Yun-Hsin Huang, Chiara Mazzuccheli, Ian D. McGreer, Bram Venemans, Fabian Walter, Simon Dye, Brad W. Lyke, Adam D. Myers, Evan Haze Nunez
DDraft version October 30, 2018
Typeset using L A TEX preprint2 style in AASTeX62
Exploring Reionization-Era Quasars III: Discovery of 16 Quasars at 6 . (cid:46) z (cid:46) . z ∼ . Feige Wang,
1, 2, 3
Jinyi Yang,
3, 2
Xiaohui Fan, Xue-Bing Wu,
2, 4
Minghao Yue, Jiang-Tao Li, Fuyan Bian, Linhua Jiang, Eduardo Ba˜nados, Jan-Torge Schindler, Joseph R. Findlay, Frederick B. Davies, Roberto Decarli, Emanuele P. Farina, Richard Green, Joseph F. Hennawi,
1, 10
Yun-Hsin Huang, Chiara Mazzuccheli, Ian D. McGreer, Bram Venemans, Fabian Walter, Simon Dye, Brad W. Lyke, Adam D. Myers, and Evan Haze Nunez
12, 13 Department of Physics, University of California, Santa Barbara, CA 93106-9530, USA Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721, USA Department of Astronomy, School of Physics, Peking University, Beijing 100871, China Department of Astronomy, University of Michigan, 311 West Hall, 1085 S. University Ave, Ann Arbor, MI,48109-1107, USA European Southern Observatory, Alonso de C´ordova 3107, Casilla 19001, Vitacura, Santiago 19, Chile The Observatories of the Carnegie Institution for Science, 813 Santa Barbara Street, Pasadena, California 91101,USA University of Wyoming, Physics & Astronomy 1000 E. University, Dept 3905 Laramie, WY 82071, USA INAF–Osservatorio di Astrofisica e Scienza dello Spazio, via Gobetti 93/3, I-40129, Bologna, Italy Max Planck Institut f¨ur Astronomie, K¨onigstuhl 17, D-69117, Heidelberg, Germany School of Physics and Astronomy, Nottingham University, University Park, Nottingham, NG7 2RD, UK Department of Physics and Astronomy, California State Polytechnic University, 3801 West Temple Ave, Pomona,CA 91768, USA Department of Physics and Astronomy, El Camino College, 16007 Crenshaw Blvd, Torrance, CA 90506, USA
ABSTRACTThis is the third paper in a series aims at finding reionzation-era quasars with the com-bination of DESI Legacy imaging Surveys (DELS) and near-infrared imaging surveys,such as the UKIRT Hemisphere Survey (UHS), as well as the
Wide-field Infrared Sur-vey Explore ( WISE ) mid-infrared survey. In this paper, we describe the updated quasarcandidate selection procedure, report the discovery of 16 quasars at 6 . (cid:46) z (cid:46) . ∼ , and present the quasar luminosity function (QLF) at z ∼ . L ) ∝ L − . in the magnitude range –27.6 < M < –25.5. We determine the quasar comoving spatial density at (cid:104) z (cid:105) =6.7 and M < − . . ± . − and find that the exponential density evolution parameter tobe k = − . ± .
18 from z ∼ z ∼ .
7, corresponding to a rapid decline by afactor of ∼ z ∼ −
5. The cosmic time between z ∼ z ∼ . Corresponding author: Feige [email protected] a r X i v : . [ a s t r o - ph . GA ] O c t Wang et al. quasar density declined by a factor of more than three within such short time requiresthat SMBHs must grow rapidly or they are less radiatively efficient at higher redshifts.We measured quasar comoving emissivity at z ∼ . z (cid:38) . (cid:38) z ∼ Keywords: galaxies: active — galaxies: high-redshift — quasars: general — cosmology:reionization INTRODUCTIONAbsorption spectra of z > z ∼ x HI (cid:38) − and becomesinsensitive to higher HI densities. If the IGM ismostly neutral, there would be appreciable ab-sorption redward of the wavelength of the Ly α emission line due to the sum of the Rayleighscattering and would give rise to long wave-length off-resonance absorptions in the form ofa damping wing profile(e.g. Miralda-Escud´e1998; Madau & Rees 2000).Despite many efforts made in the last decade,there are only three quasars currently known at z > ∼
20 quasars at z (cid:38) . (cid:38) ) sky area. Currently, the damping wing analyses havebeen only performed in the line of sight of twoknown z > α visibility(e.g.Pentericci et al. 2014) and the abundance ofhigh-redshift galaxies (e.g. Beckwith et al. 2006;Illingworth et al. 2013) and the CMB polariza-tion measurements(Planck Collaboration et al.2018), current data strongly suggest a peak ofreionization activity and emergence of the ear-liest galaxies and AGNs at 7 < z <
11 (Robert-son et al. 2015). This highlights the need toexpand the search for quasars at z (cid:38) xploring Reionization-Era Quasars z ∼ − (cid:46) M (cid:46) −
23) inthe past few years (McGreer et al. 2013; Yang etal. 2016; McGreer et al. 2018), which suggeststhat the QLF can be described with a doublepower-law function with a very steep bright-end slope of β ∼ − . α ∼ − . M ∗ ∼ −
27 . However, the pa-rameters of the z ∼ β = − . β = − .
56 when includingfainter Stripe 82 quasars. The faint-end slopeand the characteristic magnitude are even lessconstrained due to the small number of knownfaint z ∼ z ∼ α = − . M ∗ = − .
2) and a faint-end slope of β = − .
4. Nevertheless, these re-sults provide one common conclusion: the con-tribution of quasars to hydrogen reionization at z ∼ − z > . z > . ∼
300 deg area fromthe VISTA Kilo-degree Infrared Galaxy Survey(VIKING; Arnaboldi et al. 2007), and foundthat it consistent with the number density ofquasars at z ∼ k = − .
47. Howevertheir early results have large uncertainties dueto the small number of quasars and limited skycoverage. Thus, a large statistically complete and uniformly selected quasar sample at z > . Legacy Imaging Surveys (DELS; Dey et al.2018), near-infrared (NIR) surveys like UKIRTHemisphere Survey (UHS; Dye et al. 2018), aswell as the
Wide-field Infrared Survey Explore (WISE; Wright et al. 2010) mid-infrared sur-vey enables us to search for very high redshiftquasars over a large sky coverage. We presentthe discovery of a z = 6 .
63 quasar in Paper Iand a luminous z = 7 .
02 quasar in Wang et al.(2018, Paper II, hereafter).In this paper, we present the updated quasarselection procedure by further including datafrom the Pan-STARRS1 (PS1) Survey (Cham-bers et al. 2016). We report the discovery of 16quasars at 6 . (cid:46) z (cid:46) . z ∼
6, and present the QLF at z ∼ . § § § § z ∼ . § §
7. Finally, wereport the discovery of five additional z ∼ http://desi.lbl.gov/ Wang et al.
Table 1.
Photometric Information of Datasets Used in This Paper
Survey Band Depth (5- σ ) Sky Area (deg ) AB offset RefDELS (DR4+DR5) a g , r , z g ps , r ps , i ps , z ps , y ps J J J J W b Note —All magnitudes in this table are in AB system. The conversion factors from VEGA to AB are listed in the fifth column. a The depth for DELS is for those area with only one photometric pass. b This value was estimated using the magnitude-error relation from Yang et al. (2016), assuming the number of coverage equals to 24, whichcorresponds to the mean coverage of ALLWISE dataset. dard ΛCDM cosmology with Hubble constant H = 70 km s − Mpc − , and density parametersΩ M = 0 . Λ = 0 . QUASAR CANDIDATES SELECTIONAt z ∼
7, the Ly α emission line in the quasarspectrum redshifts to ∼ . µ m. Thus, quasarsat z ∼ z − J color due to the presence of neutral hy-drogen at high redshift that absorbs most of theemission blueward of the Ly α emission line inquasar spectra. As a result, we need both deepoptical and NIR photometry to select quasarsat z ∼
7. 2.1.
Imaging Data
For optical bands, we mainly used data fromDELS , which consists of three different imagingsurveys: the Dark Energy Camera Legacy Sur-vey (DECaLS), the Beijing-Arizona Sky Sur-vey (BASS; Zou et al. 2017) and the Mayall z -band Legacy Survey (MzLS). These three sur-veys jointly image ∼ of the extra-galactic sky visible from the northern hemi-sphere in three optical bands ( g , r and z ). TheDECaLS survey covers ∼ extragalac-tic sky with Decl. ≤ ◦ and the BASS+MzLScover ∼ sky with Decl. ≥ ◦ . There http://legacysurvey.org/ is a total of ∼
300 deg regions at 32 . ◦ < Decl . < . ◦ where DECaLS overlaps withBASS+MzLS. An overview of DELS surveyscan be found in Dey et al. (2018). We alsoinclude PS1 photometric data in our selection,which provides a 3 π sky coverage in g ps -, r ps -, i ps -, z ps -, and y ps -bands. The PS1 photo-metric data was obtained from MAST CasjobsPS1 Archive . Although the PS1 survey is shal-lower than DELS, it has additional y ps -bandwhich is redder and narrower than the reddest z -band of DELS. At NIR, we combine UHSDR1 with public data from UKIRT InfraredDeep Sky Survey (UKIDSS; Lawrence et al.2007) DR10, VISTA Hemisphere Survey (VHS;McMahon et al. 2013) DR5 and VIKING DR4.The UHS and UKIDSS data were obtained fromWFCAM Science Archive , while the VHS dataand VIKING data were obtained from VISTAScience Archive . In addition, we used the ALL-WISE release of the WISE data, which com-bines the original WISE survey (Wright et al.2010) with data from the NEOWISE (Mainzeret al. 2011) post-cryogenic phase. Because UHSDR1 only provides J -band photometry, we only http://mastweb.stsci.edu/ps1casjobs/ http://wsa.roe.ac.uk/index.html http://horus.roe.ac.uk/vsa/ http://wise2.ipac.caltech.edu/docs/release/allwise/ xploring Reionization-Era Quasars J -band in our selection in order to havea homogeneous selection procedure over thewhole DELS footprint. The basic characteris-tics of these imaging surveys are listed in Table1. 2.2. z ∼ Quasar Candidate Selection
We started from DELS data release 4 (DR4,MzLS+BASS data) and data release 5 (DR5,DECaLS data), which contains ∼ . ∼ . z -band detection but are not ob-served/not detected (at 5- σ level) in g and/or r bands. This results in a total of ∼ . . (cid:48)(cid:48) g ps , r ps and i ps bandsat 5 σ levels. (3) We then select targets whichhave both DELS z and PS1 y ps -band withat least 7- σ detections, have PS1 y ps -bandbrighter than 21.5 and fainter than 15 mag-nitude, and have PS1 z ps -band undetected in5- σ or with PS1 z ps − y ps > .
5. (4) We per-formed forced photometry on PS1 images usingan aperture radius of a 1 . (cid:48)(cid:48) . (cid:48)(cid:48) i ps ,forced brighter than23.1 or z ps , forced − y ps , forced < .
7. (5) We fur-ther reject targets identified as extended sourcesin both PS1 photometry (magnitude differencesbetween aperture photometry and PSF pho-tometry larger than 0.3 ) and DELS photome-try (type not equals “PSF”). Considering theastrometry uncertainties in both PS1 and DELSare relative small (i.e. (cid:46) . (cid:48)(cid:48) . (cid:48)(cid:48)
0. (6) We cross-matchedour candidates with infrared photometric cat-alogs from UHS, UKIDSS, VHS, and VIKINGand rejected those targets with y ps − J > . y ps1 −J J − W M L T
OtherSelectedPaper I+IINew
Figure 1.
The y ps ,AB − J V EGA vs. J V EGA − W V EGA color-color diagram. The cyan line andcyan filled circles represent the color-redshift rela-tion predicted using simulated quasars (McGreer etal. 2013; Yang et al. 2016) from z = 6 . z = 7 . z = 0 .
1. The large cyan circles high-light the colors at z = 6.0, 6.5, 7.0, and 7.2. Theorange open asterisks denote two z > . z (cid:38) . z > . z > . z > y ps from forced photom-etry on PS1 images. The steel blue crosses, greenopen triangle, and magenta open squares depict thepositions of M, L, and T dwarfs, respectively (Kirk-patrick et al. 2011; Best et al. 2015). or J − W < . J -band.Figure 1 shows the y ps − J/J − W g ps , r ps , i ps and DELS g , r bands, affected by cosmic rays,or contaminated by nearby bright stars are re-moved. Wang et al.
The selection procedure described aboveyields a total of 121 quasar candidates left forspectroscopically follow-up observations. Werefer these dropouts as our main quasar candi-dates in the following sections. Our criteria fortargeting z ∼ S/N ( g, r, g ps , r ps , i ps ) < . S/N ( z, y ps ) > . z > , < y ps < . S/N ( z ps ) < . or z ps − y ps > . i ps , forced > . , z ps , forced − y ps , forced > . y ps − J < . J − W > . Supplementary Quasar Selection
Recently, an ultra-luminous (i.e. M < −
29) quasar population has been discoveredat high redshift (Wang et al. 2015; Wu et al.2015; Wang et al. 2016). Such ultra-luminousquasars are detected in the dropout bands dueto their extreme brightness. Our selection pro-cedure presented in § z . On the other hand,strong gravitationally lensed quasars would alsobe missed by requiring non-detections in bluerbands as the lensing galaxy would contributeflux in those bands (e.g. McGreer et al. 2010).In order to recover such quasar population at z (cid:38) . g ps , r ps , i ps ) − y ps > . , z ps − y ps > .
5) atgalactic latitude greater than 5 degrees. Thenwe cross-matched this catalog with all availableNIR photometry data (UHS, VHS, VIKING,2MASS) and the ALLWISE photometry cata-log. We then used the y ps − J/J − W W − W > . SPECTROSCOPIC OBSERVATIONSWe obtained spectroscopic follow-up obser-vations of the quasar candidates with theMMT/Red Channel spectrograph (Schmidt etal. 1989), MMT/MMIRS (McLeod et al. 2012),MMT/ Binospec (Fabricant et al. 1998), Magel-lan/FIRE (Simcoe et al. 2008), Magellan/LDSS3-C (Stevenson et al. 2016), KECK/DEIMOS(Faber et al. 2003), LBT/MODS (Byard &O’Brien 2000), Gemini/GMOS (Hook et al.2004) and P200/DBSP (Oke & Gunn 1982),over ∼
25 observing nights from April 2016 toJuly 2018.We observed 33 candidates with MMT 6.5m telescope, with 26 main candidates targetedwith Red Channel Spectrograph, 5 main candi-dates targeted with MMIRS spectrograph, onemain candidate and one supplementary candi-date observed with Binospec. We used the 270l mm − grating on Red Channel spectrographcentered at 9000 ˚A, providing wavelength cov-erage from 7200 ˚A to 10800 ˚A. We used the1 . (cid:48)(cid:48) . (cid:48)(cid:48) R ∼
640 and R ∼ µ m anda resolution of R ∼ ,
000 with a 1 . (cid:48)(cid:48) (cid:48) × (cid:48) fields of view. We used 270 gratingcentered at 7400 ˚A with a 1 . (cid:48)(cid:48) R ∼ xploring Reionization-Era Quasars Table 2.
Observational information of 16 new quasars reported in this paper.
Name Telescope Instrument Exposure (sec) OBS-DATE (UT)DELS J041128.63 − − a MMT Red Channel 6000 20180120, 20180203DELS J110421.59+213428.8 Magellan/LBT FIRE/MODS b × c Hale 200 inch DBSP 15900 20160908, 20160909, 20160911VHS J210219.22 − a This quasar was discovered by Matsuoka et al. (2018b) independently. b We used binocular mode by using both MODS1 and MODS2. c This quasar was discovered by Mazzucchelli et al. (2017) independently. throughput mode which provides a resolutionof R ∼ . (cid:48)(cid:48) R ∼ . (cid:48)(cid:48) R ∼ µ m. Two main candidateswere observed with Magellan/LDSS3-C usingthe VPH red grating with 1 . (cid:48)(cid:48) R ∼ µ m. We observed three supplementary candi-dates using KECK/DEIMOS with 830G grat-ing.In total, we spectroscopically observed 65main candidates and 11 supplementary can- didates. The data obtained from Red Channel,Binospec, MODS and DBSP were reduced us-ing standard IRAF routines. The GMOS datawere reduced using the Gemini IRAF packages.The data obtained with FIRE and MMIRS werereduced using a custom set of Python routines(Wang et al. 2017) which includes dark subtrac-tion, flat fielding, sky subtraction, wavelengthcalibration and flux calibration. The data ob-tained with DEIMOS were reduced using theXIDL suite of astronomical routines in theInteractive Data Language (IDL), which wasdeveloped by X. Prochaska and J. Hennawi. RESULTS4.1.
Discovery of 16 New Quasars at . (cid:46) z (cid:46) . ∼ xavier/IDL/ Wang et al. f λ ( − e r g s − c m − Å − ) J0923+0402, z=6.61 8000 8500 9000 9500 10000012 f λ ( − e r g s − c m − Å − ) J0910-0414, z=6.638000 8500 9000 9500 10000012 J2102-1458, z=6.648 8000 8500 9000 9500 10000012 J1216+4519, z=6.6548000 8500 9000 9500 1000005 J0837+4929, z=6.71 8000 8500 9000 9500 100000.02.55.0 J0910+1656, z=6.728000 8500 9000 9500 10000024 J1104+2134, z=6.74 8000 8500 9000 9500 1000002 J0829+4117, z=6.7688000 8500 9000 9500 10000
Wavelength (Å)
Wavelength (Å)
05 J0839+3900, z=6.905
Figure 2.
Spectra of the 16 newly discovered z (cid:38) . β , Ly α , and N v from left to right. The grey lines denote 1 σ flux error vector. All spectra are binned into 10 ˚A in wavelengthspace using 1/ σ weighted mean algorithm. xploring Reionization-Era Quasars . InPaper I and Paper II, we have reported twoquasars from the main sample, DELS J1048–0109 at z = 6 . z = 7 .
02. In Fan et al. (2018), we reportedone strong gravitationally lensed quasar (UHSJ0439+1634) at z = 6 .
511 from the supplemen-tary sample, including followup
HST imagingand detailed lens model results. All redshiftsmeasured from (sub)mm emission lines fromhost galaxy have four digits, redshifts measuredfrom Mg ii broad emission lines have three dig-its, and redshifts measured from Ly α have twodigits throughout the paper.Here, we report additional 16 new quasars at6 . (cid:46) z (cid:46) .
0. Fifteen of them come fromour main sample, and one object, J210219.22–145854.0, is from our supplementary sample.J2102–1458 is not covered by our main selectionbecause it is outside the DELS footprint andthus not included in our QLF measurementsin the following sections. The details of thespectroscopic observations of these newly dis-covered quasars are listed in Table 2. Figure2 shows the discovery spectra of these quasars.For those quasars with spectra available onlyin optical, we measured quasar redshifts fromLy α and N v lines by fitting the observed spec-tra to the SDSS quasar template (Vanden Berket al. 2001) using a visual recognition assistantfor quasar spectra software (ASERA; Yuan etal. 2013). The typical redshift errors are about0.03 due to the combination of low spectral reso-lution and strong absorptions blueward of Ly α .We have already obtained near-IR spectra forsome of these quasars, for which we estimate theredshifts by fitting Mg ii emission lines (Yang J. in preparation ). The redshift uncertainties for The list of non-quasars is available from the corre-sponding author on reasonable request. quasars that have near-IR spectra are usuallyaround 0.01. Note that the redshift uncertain-ties quoted here do not take the possible shiftcompared with (sub)-millimeter emission linesfrom quasar host galaxies, which could be upto several thousand kilo-meter per seconds (e.g.Decarli et al. 2018). We listed redshifts andphotometric information of newly discoveredquasars in Table 3. There are four quasars thatdo not have any available NIR photometry. Forthese objects, we obtained additional J -bandphotometry with UKIRT/WFCAM (Casali etal. 2007).We use the continuum magnitude at rest-frame 1450 ˚A in the determination of the QLF.At z > .
3, the rest-frame 1450 ˚A is red shiftedto observed wavelengths longer than 1.06 µ m,which is beyond the useful wavelength coverageof our optical spectra. Thus it is not possible toestimate the 1450 ˚A magnitudes by directly fit-ting power-law to the discovery spectra shownin Figure 2. Instead, we scale the compositespectra of luminous low redshift quasars (Sels-ing et al. 2016) to the Galactic extinction cor-rected J-band photometry of each quasar. Thenwe estimate the 1450 ˚A magnitudes from thescaled composite spectrum. In table 3, we listthe apparent and absolute AB magnitudes atrest-frame 1450 ˚A in Column 3 ( m ) and Col-umn 4 ( M ), respectively.Figure 3 shows the redshift and M distri-bution of all quasars at z ≥ . M (cid:38) − . z > . M < − . ∼
10 previously known quasars were discoveredfrom multiple quasar surveys (Mortlock et al.2011; Venemans et al. 2013, 2015; Tang et al.0
Wang et al.
Table 3.
Properties of 16 new z (cid:38) . Name Redshift m M z DELS , AB y ps1 , AB J VEGA W VEGA
NIR Survey QLFDELS J083946.88+390011.5 6.905 ± c ± ± ± ± ± ± − ± ± ± ± ± ± ± d YDELS J082931.97+411740.4 6.768 ± c ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± d YDELS J083737.84+492900.4 6.710 ± c ± ± ± ± ± ± ± c ± ± ± ± ± ± d YVHS J210219.22 − ± c ± ± ± ± ± d NDELS J091054.53 − ± ± ± ± ± ± ± a ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± b ± c ± ± ± ± ± ± ± ± ± ± ± ± ± ± c ± ± ± ± ± ± ± ± ± ± ± ± ± Note:
The column “NIR Survey” indicate where the J band photometry comes from. The last column indicates whether a quasar is used for QLFmeasurements. a This quasar was discovered by Matsuoka et al. (2018b) independently. b This quasar was discovered by Mazzucchelli et al. (2017) independently. c The redshift measured from near-infrared spectra by fitting broad Mg ii emission lines (Yang, J., et al. 2018 in preparation ). d The J-band photometry were obtained using UKIRT/WFCam (Programs: U/17B/UA01 and U/17B/D04). z > . Notes on Individual Quasars
DELS J083946.88+390011.5 . J0839+3900is a broad absorption line (BAL) quasar at z = 6 .
905 with strong blue-shifted N v absorp-tion. We obtained deep Gemini/GNIRS spec-trum which shows that J0839+3900 is a Low-ionization BAL (LoBAL) quasar with strongblue shifted Mg ii absorptions. It is the highestredshift known LoBAL quasar. The NIR spec-trum and related physical parameter measure-ments will be reported in the future together with NIR spectra of other z > . in preparation ). DELS J092347.12+040254.4 . J0923+0402is a BAL quasar. It was independently discov-ered by Matsuoka et al. (2018b).
DELS J091054.53–041406.8 . J0910–0414is a BAL quasar. It was initially identified as a z ∼ . z = 6 . v is absorbed. DELS J070626.39+292105.5 . J0706+2921is the most luminous z > . M =–27.51, and isabout 0.3 magnitude brighter than the previousrecord holder (Venemans et al. 2015). xploring Reionization-Era Quasars Redshift −30−29−28−27−26−25−24−23 M Å OtherSelectedJ0439Paper I+IINew
Figure 3.
The redshift and absolute magnitudedistribution of z ≥ z > z = 6 . DELS J162911.29+240739.6 . J1629+2407was independently discovered and reported byMazzucchelli et al. (2017).
DELS J153532.87 + . J1535+1943has a very red y ps − J color of ∼ ∼ .
0) at similar red-shifts. The low SNR spectrum shown in Figure2 shows that the break at blueward of Ly α is notas sharp as others, which suggests J1535+1943might be a reddened quasar or has a proximatedamped Lyman-alpha (PDLA) system in frontof it. DELS J131608.14+102832.8 . J1316+1028is a BAL quasar at z = 6 .
35 with strong blue-shifted N v absorption.4.3. BAL Quasar Fraction
BAL quasars show gaseous outflows whichcause strong blue-shifted absorptions in quasarspectra. Previous studies based on spectralanalyses indicated that observed BAL quasarscomprise about ∼
15% of the quasar populationat low and intermediate redshifts, without sig-nificant redshift dependence (e.g. Reichard et al.2003; Hewett & Foltz 2003; Knigge et al. 2008;Gibson et al. 2009). However, Allen et al. (2011)found a strong redshift dependence of the BALquasar fraction with a factor of 3 . ± . z ∼ .
0, down to z ∼ .
0. The redshift de-pendence implies that orientation effect alone isnot sufficient to explain this trend.An alternative model which allows cosmic evo-lution of the BAL quasar fraction is that radia-tion driven winds are the likely origin of quasaroutflows (e.g. Risaliti & Elvis 2010). The ra-diation driven winds can be generated by thequasar accretion disc under a variety of physicalconditions, and the possibility of their existenceis a function of physical parameters such as theBH mass, Eddington ratio, and X-ray to UVflux ratio (e.g. Risaliti & Elvis 2010). Thus, in-vestigating whether the BAL quasar fraction isdifferent at the EoR would help us understandthe nature of BAL quasars and probe whether itis only related to the orientation or also affectedby other physical parameters.Four newly discovered quasars in our sam-ple that show strong blue-shifted N v absorp-tions, indicating that they are BAL quasars,J1316+1028 at z = 6 .
35, J0923+0402 at z =6 .
61, J0910–0414 at z = 6 .
63, and J0839+3900at z = 6 . iv broad absorption troughs (Paper II). Weonly estimate the BAL fraction in our mainquasar sample because it is statistically com-2 Wang et al. plete. There are five published PS1 quasarsthat also satisfy our main selection procedure(See Table 4). Thus, we need to include thesefive PS1 quasars, J1048–0109 reported in PaperI and J0038–1527 reported in Paper II, but ex-clude J2102–1458 when counting the BAL frac-tion. We visually inspected the spectra of fivePS1 quasars (Venemans et al. 2015; Mazzuc-chelli et al. 2017), only PSO J036+03 showssmall possible absorption troughs at the blue-ward of S iv and C iv emission lines. However,the spectrum of PSO J036+03 in Venemans etal. (2015) shows several unusual bumps, whichcould be due to flux calibration issues typicalof Echelle spectra. We treat PSO J036+03 as apossible BAL quasar. Therefore, the observedBAL quasar faction in our main quasar sampleis 5(6)/23=21.7(26.1)%, which is slightly higherthan that at lower redshift (e.g. Reichard et al.2003; Hewett & Foltz 2003; Knigge et al. 2008;Gibson et al. 2009). Note that the spectra ofsome quasars do not cover the C iv lines and wedo not know whether they are real Non-BALquasars. In addition, the color selection biaswould underestimate the BAL quasar fractionby a few percents (Reichard et al. 2003) due tothat BAL quasars usually have slightly reddercolors. Thus, the observed BAL quasar fractiongiven here should be treated as a lower limit.We will revisit this question in details after col-lecting NIR spectra for all quasars.4.4. Radio Properties
Bright radio sources at the EoR allow us tostudy the cosmic reionization by detecting 21cm absorptions from intervening neutral IGM(e.g. Carilli et al. 2007; Semelin 2016). In addi-tion, powerful radio jets play a key role in theformation and build-up of SMBHs (e.g. Volon-teri et al. 2015). The radio loud fraction ofquasars is found to be ∼
10% from low redshiftsup to z ∼ z > .
5, where the universe
Redshift −28−27−26−25−24−23 M R=500R=100R=30
Figure 4.
The redshift and absolute magni-tude distribution and radio loudness constraints of z > R = 500, R = 100, and R = 30, respectively.It shows that both FIRST and NVSS surveys aretoo shallow to rule out those z > . R ∼ > is relatively neutral. The large quasar samplepresented here allows investigation of the radioproperties of early quasars. We cross-matchedall known z (cid:38) . z > . R = f /f (e.g. Stocke etal. 1992; Jiang et al. 2007), where f and f are flux densities at rest-fame 6cm and 2500 ˚A xploring Reionization-Era Quasars Table 4.
Previously known z (cid:38) . Name Redshift m M z DELS , AB y ps1 , AB J VEGA W VEGA
Ref. a QLFDELS J003836.10 − ± ± ± ± ± ± ± − ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a These objects were discovered by several studies: (1) Paper II; (2) Paper I; (3) Venemans et al. (2015); (4) Mazzucchelli et al. (2017).
Figure 5.
Stacked VLA FIRST image of 17 z > . ∼ × − mJy/beam,four times better than FIRST depth. +2 σ contoursare shown as cyan solid lines and –2 σ contours areshown as cyan dashed lines. No signal is detectedat the quasar position. respectively. We estimate the f from m by assuming f ν ∝ ν − . (Lusso et al. 2015), andestimate f from the 1.4GHz observed fluxdensity by assuming f ν ∝ ν − . (e.g. Wang etal. 2007). Figure 4 shows the constraints onradio loudness of z > . R (cid:38) (cid:46) σ level. Future deeper radioimaging of these z > . z > . QUASAR LUMINOSITY FUNCTION AT Z ∼ . A Complete Quasar Sample
Since most of our discoveries are from ourmain sample, we will calculate the QLF onlyusing the main quasar sample. There are 15quasars reported in this paper from the mainsample, two quasars reported in Paper I andPaper II, and another 5 quasars discovered byPS1 high redshift quasar survey (Venemans etal. 2015; Mazzucchelli et al. 2017) that alsosatisfy our selection criteria. The previouslyknown z (cid:38) . g, r, i bands (Eq. 1)and z -dropout cut (Eq. 5) in our selection pro-cedure, our selection is highly incomplete at4 Wang et al. z (cid:46) . z < .
45 from Mazzucchelli etal. (2017) when calculating the QLF. Our mainsample missed four z > . . 05 and − . < M < − . 5. Theredshifts of these quasars were measured fromLy α and Mg ii broad emission lines or from [C ii ]emission lines in sub-millimeter (Decarli et al.2018) with a median redshift of (cid:104) z (cid:105) ∼ . 7. Theapparent and absolute AB magnitudes of con-tinuum at rest-frame 1450˚A ( m and M )for those recovered known quasars are derivedusing the same method described in Section 4.1.5.2. Area Coverage Since the PS1 covers the whole DELS foot-print and we do not reject any sky area that isnot covered by the NIR surveys (i.e. a smallarea at Decl . > ◦ ), the total searching area isbasically the footprint of DELS DR4 and DR5.We require quasar candidates to have at leastone DELS z -band observation. But we do notlimit our selections with g and/or r bands ob-servations, thus the g - and r -bands observationsdo not affect our sky coverage estimate. TheDELS DR4 covers 3,267 deg in z -band andthe DELS DR5 covers 9972 deg in z -band .However, there are ∼ 200 deg overlap regionsobserved, thus we need to avoid double-countingthe overlap regions when estimating the areacoverage. Instead of adding the DR4 and DR5coverage, we generate a photometric catalog in- http://legacysurvey.org/dr4/description/ http://legacysurvey.org/dr5/description/ cluding both DELS DR4 and DR5 photometricdata and only keep one object if it has dupli-cated detections in DR4 and DR5. We then usethe Hierarchical Equal Area isoLatitude Pix-elization (HEALPix; G´orski et al. 2005) to es-timate the sky coverage of DR4+DR5 followingJiang et al. (2016). The final estimated sky areaby HEALPix is 13,020 deg , which is consistentwith the area estimated by adding DR4 andDR5 coverage and removing overlap regions.5.3. Selection Function Following our previous works (McGreer et al.2013, 2018; Yang et al. 2016; Jiang et al. 2016),we use simulations to estimate the completenessof our selection procedure, including the colorcuts and flux limits that we applied in Section2.2. The simulation is performed under the as-sumption that the shape of quasar SEDs doesnot evolve with redshift. We generate a gridof model quasars using the simulations by Yanget al. (2016), which is an updated version ofthe simulations by McGreer et al. (2013). Themodeled quasar spectra are designed to matchthe colors of ∼ . < z < . α forest. Finally, photometry is derived fromsimulated quasars and photometric errors areadded for each survey by matching the observedmagnitude and error relations with a large rep-resentative point source sample (Yang et al.2016). The PS1 coverage depends on the skyposition and the depth is not uniform. Cham- xploring Reionization-Era Quasars Redshift −27.5−27.0−26.5−26.0−25.5 M Figure 6. Selection function of our z ∼ M , z ) bin.The orange solid asterisks denote newly discoveredquasars reported in this work and blue open aster-isks are previously known z > . 45 quasars. Thecontours are selection probabilities from 0.7 to 0.1with an interval of 0.2. bers et al. (2016) gives the all-sky distributionof magnitude limits for 50% and 98% complete-ness on the PS1 3 π stacked data, which indi-cates that PS1 DR1 catalog is ∼ 50% completefor y ps (cid:38) . (cid:38) 98% completefor y ps < 21 objects (except for some low galac-tic latitude regions). We need to consider thesky position-dependent photometric uncertain-ties caused by the PS1 inhomogeneous cover-age. To correct this effect in our simulations, wemapped the PS1 spatial surveying depth withinour searching area and fit a 2D magnitude–coverage–error relation following the procedureexplored by Yang et al. (2016). Similarly, wealso apply the same method for the simulatedDELS photometry. We refer to McGreer et al.(2013) and Yang et al. (2016) for more detaileddescriptions of the simulation.We use the simulation described above to es-timate the completeness of our selection crite-ria. To derive a selection function, we construct S e l e c t i o n F un c t i o n BrightModerateFaintAll Figure 7. Selection function as a function ofredshift for three different luminosity bins. Theblue, orange and green dashed lines denote selectionfunction for quasars with − . < M < − . − . < M < − . 2, and − . < M < − . 5, respectively. The red solid line representsthe selection function over the whole magnituderange. The shaded region shows the redshift binwe used for calculating QLF. a grid of simulated quasars distributed evenlyin ( M , z ) space with 100 quasars bin − of∆ M =0.1 and ∆ z =0.05. Then we computethe average selection probability, p ( M , z ),in each ( M , z ) bin. The computed selec-tion function in the ( M , z ) space as well asthe seventeen quasars we used for QLF mea-surement are shown in Figure 6. We notethat there are two faint quasars have proba-bilities below 30%: J0910+1656 at z = 6 . z = 6 . ∼ α emission and thusis bright in both y ps1 and z DELS , although it isfaint in rest-frame 1450˚A. J1216+4519 is notthat extreme but has similar situation withJ0910+1656. J1216+4519 is bright in y ps1 and z DELS but faint in M . There is no any quasarfound in the highly complete region at redshiftbetween 6.6 and 6.9, which is probably because6 Wang et al. there is no such bright quasar in our searchingarea at this redshift range. Not surprising, thisis because the number density of such luminousquasar at z > . z > . M brighter than –27 previously known at z > . M from –27.51 to –25.51. In order to include a sta-tistical quasar sample in each magnitude bin, wedivide our sample into 3 magnitude bins with∆ M =0.7 mag over the magnitude range –25.5 < M < –27.6. Figure 7 shows the selec-tion function as a function of redshift in threedifferent luminosity bins. As expected from the z -dropout cut, our main selection procedure hasa very sharp change of completeness at z ∼ . y ps and DELS z -band, our main selectionlimits the quasar redshift to be lower than 7 . z (cid:38) . 8. For these reasons, we choose the red-shift range from 6.45 to 7.05 (the shaded re-gion in Figure 7) when calculating the QLF. Asshown in Figure 6, the main selection proceduremisses very bright quasars at z (cid:46) . i -band.5.4. Spectroscopic Completeness As mentioned in Section 3, we spectroscopi-cally observed 65 main candidates. These 65observed targets include quasars J1048–0109,and J0038–1527, which have been published inPaper I and Paper II. There are four candi-dates in our spectroscopically observed samplethat can neither be rejected nor confirmed ashigh redshift quasars based on available spec-tra, and we can not count these four can-didates as spectroscopically observed targets.Thus, the overall success rate of our main se-lection is 29.5% (18/61). The success rateat y ps < . y ps1 (AB mag) N u m b e r QSOsNon-QSOsUnobserved 0.00.20.40.60.81.0 Sp e c . C o m p l e t e n e ss Spec. Completeness Figure 8. Spectroscopic completeness of our mainsample. The magenta dashed line denotes the spec-troscopic completeness as a function of y ps -bandmagnitude. The histogram is divided into sev-eral components filled by different colors and rep-resents newly identified high-redshift quasars (or-ange), non-quasars (blue), and unobserved can-didates (gray). The black dotted line represents y ps =21.0 mag, which we treat as our quasar sur-vey flux limit. and declines rapidly towards fainter objectsbecause the dropout bands are not deepenough for fainter candidates. There are fivemore previously known quasars, J0226+0302,J1212+0505, J1724+1901, J2132+1217, andJ2232+2930 that can also be treated as spec-troscopically observed targets. So, 66 out of121 main sample candidates were spectroscop-ically observed in total. The y ps magnitudedistribution of our observed and unobservedcandidates is shown in Figure 8. The numberof spectroscopically observed candidates is afunction of y ps -band magnitude, which is usedto correct the incompleteness by assuming theprobability of an unobserved candidate to be aquasar is the same as in the observed sampleat a certain magnitude. As shown in Figure 8,we do not identify any high redshift quasar at y ps > . 0, which means we can not correctthe quasar fraction for unobserved candidates xploring Reionization-Era Quasars −28.0−27.5−27.0−26.5−26.0−25.5−25.0 M −10 −9 Φ ( M ) [ M p c − m a − ] z=6 QLF (Jiang+2016)z=6 QLF (Willott+10)Φ ∝ M −2.35 DPL (β = -2.54)Binned QLF Figure 9. Binned quasar luminosity function. Theorange solid symbols represent our newly measuredbinned QLF at z ∼ . 7. The blue dashed line isour best fit with β = − . ± . 22. The magentasolid line denotes double power-law fit. The blackdotted and dash-dotted lines represent z ∼ z ∼ . z ∼ at this magnitude range. Figure 8 shows thatthe number of main candidates drops very fastat y ps > . 0, which is caused by the fact thatthe depth of PS1 is not uniform and faint ob-jects can only be detected at 7- σ level in deepregions. Considering these two limitations, weonly use the y ps < . Luminosity Function at z ∼ . V a method (Page & Carrera 2000),after all incompleteness corrections have beenapplied for each quasar. The binned QLF is shown in Figure 9. TheQLF can be well characterized by a singlepower-law, Φ( L ) ∝ L β , or,Φ( M ) = Φ ∗ − . β +1)( M +26) , (8)where we only consider luminosity dependencebut ignore redshift evolution over our narrowredshift range. The best fits are Φ ∗ = (6 . ± . × − Mpc − mag − and β = − . ± . M, z ) = Φ ∗ ( z )10 . α +1)( M − M ∗ ) + 10 . β +1)( M − M ∗ ) , (9)where α and β are the faint-end and the bright-end slopes, M ∗ is the characteristic magnitudeand Φ ∗ ( z ) = Φ ∗ ( z = 6) × k ( z − is the nor-malization. Since our binned QLF only coversa narrow luminosity range, we can not fit α and M ∗ ( z ). Currently, there is no z > . α and M ∗ to the z ∼ α = − . 90 and M ∗ = − . 2. Here weuse least-square fitting rather than maximumlikelihood fitting due to the lack of faint quasarsand unknown faint-end slope and characteris-tic magnitude. Our least-square fitting givesΦ ∗ = (3 . ± . × − Mpc − mag − and β = − . ± . z (cid:46) β ≤ − α ∼ − . − − . β = − . ± . z ∼ M = − . β = − . ± . 16, if they fit asingle power-law to all SDSS quasars ( − . ≤ M ≤ − . Wang et al. slope β = − . +0 . − . at z ∼ 6. The large differ-ence in the bright-end slope between Kulkarniet al. (2018) and previous works is mainly be-cause Kulkarni et al. (2018) measured a verybright characteristic magnitude which is M ∗ = − . 21. In other words, the QLF measured byKulkarni et al. (2018) follows a single power-law with a slope of − . +0 . − . at M (cid:38) − . − . ≤ M ≤ − . β = − . ± . 22 at z = 6 . − . < M < − . β = − . ± . 29 if wefix the faint-end slope and characteristic mag-nitude to be the values derived by Jiang et al.(2016), it changes to β = − . ± . 22 if we fixthe bright end slope and characteristic magni-tude to be the values derived by Kulkarni et al.(2018). Our result suggests that the QLF slopedoes not evolve strongly from z ∼ z = 6 . − . < M < − . DISCUSSION6.1. Density Evolution of High-RedshiftLuminous Quasars A rapid decline in the comoving number den-sity of luminous quasars at high redshift wassuggested Fan et al. (2001), who fit an expo-nential decline to the quasar spatial density, ρ ( 47. Thisvalue has been frequently used in many previ-ous works (e.g. Willott et al. 2010; Kashikawaet al. 2015). An even more rapid decline in thecomoving number density from z ∼ z ∼ k ∼ − . 7) is claimed by recent studies (Mc-Greer et al. 2013; Jiang et al. 2016). Here, weexplore in detail the spatial density evolutionof luminous quasars at higher redshifts. Thespatial density of quasars brighter than a givenmagnitude M can be calculated by integrating Redshift ( z , M < ) [ G p c ] k = 0.78Richards+2006Fan+2001Kashikawa+2015Willott+2010Richards+2006McGreer+2013Yang+2016Jiang+2016This Work10.0 5.0 3.0 2.0 1.5 1.2 1.0 0.8 0.7 Universe Age (Gyr) Figure 10. Density evolution of luminous quasars.The grey solid line and dashed line denote evolu-tion model from Richards et al. (2006) and Fan etal. (2001), respectively. The black dashed line de-notes density evolution model from z ∼ z ∼ . k = − . 78. The grey solid line and grey dot-dashed lines are from Richards et al. (2006) andFan et al. (2001), respectively. The orange solidcircle denote our measurement at the highest red-shift. The blue open circles are density measuredfrom the binned SDSS quasar luminosity function(Richards et al. 2006). The magenta and red opensquares are density measured using binned luminos-ity function from McGreer et al. (2013) and Jianget al. (2016), respectively. The cyan, steel blue andyellow open diamonds denote densities integratedfrom QLF measured by Yang et al. (2016), Willottet al. (2010) and Kashikawa et al. (2015), respec-tively. the QLF: ρ ( < M, z ) = (cid:90) M −∞ Φ( M, z ) dM , (10)We can also estimate the density using the1/ V a method based on individual quasars andselection function, V a = (cid:90) ∆ z p ( M , z ) dVdz dz , (11)where p ( M , z ) is the selection function ateach magnitude and redshift bin. The total spa- xploring Reionization-Era Quasars ρ ( < M, z ) = (cid:88) i V ia , σ ( ρ ) = (cid:34)(cid:88) i (cid:18) V ia (cid:19) (cid:35) , (12)We estimate the density of quasars brighterthan M < –26 at z ∼ . . ± . 11) Gpc − mag − , by summing over all thequasars used for QLF measurement and with M < –26 using Eq. 12. With the samemethod, Jiang et al. (2016) measured the den-sity at z = 6 to be (1 . ± . 33) Gpc − usinga large sample of luminous SDSS quasars. Thespatial density of quasars at z = 6 . z = 6. InFigure 10, we show the estimated quasar spatialdensity at z ∼ . 7, together with the results at z < z ∼ z ∼ k = − . ± . 18 from z ∼ z ∼ . z ∼ . z ∼ z ∼ ∼ z ∼ − 5. If such declineextends to higher redshift, we expect to see onlyone such luminous quasar over the whole visiblesky at z ∼ 9. This means that we are finally wit-nessing the first quasars in the EoR and we willbe badly limited by the small number of such quasars when studying the reionization historyand SMBH growth history.Quasar evolution at z > e -folding times available for BH ac-cretion. The rapid decline of luminous quasarspatial density within such short cosmic time(i.e., ∼ 121 Myrs, or three e -folding times) re-quires that SMBHs must grow rapidly from z ∼ . z ∼ z ∼ . 7. For J0706+2921, the bright-est quasar in our sample, it takes 20 e -foldingtimes, or the age of the universe at z ∼ . 6, togrow from a 10 M (cid:12) stellar black hole, assumingthe radiation efficiency (cid:15) =0.1. The existence ofthese luminous quasars helps determine whetherstandard models of radiatively efficient accre-tion from stellar seeds are still allowed, or al-ternative models of BH seed formation and BHaccretion (super-Eddington or radiatively ineffi-cient) are required (e.g. Volonteri & Rees 2006).In addition, the determination of luminousquasar spatial density evolution at high redshifthas important consequences in understandingearly BH growth and BH-galaxy co-evolution(e.g. Wyithe & Loeb 2003; Hopkins et al. 2005;Shankar et al. 2010). Combining the dark mat-ter halo mass and duty circle inferred fromquasar clustering measurements (Shen et al.2007), Shankar et al. (2010) predict the QLFat z > z (cid:38) z = 3 − z > . M halo (cid:38) M (cid:12) ), less numerous dark mat-ter halos than that of luminous z ∼ Wang et al. z [ e r g s H z M p c ] Khaire+2015Madau+2015Haard+2012Khaire+2015Re-calculationsDPL ( = 1.90, = 2.54)Single PL ( = = 2.35) Universe Age (Gyr) Figure 11. The quasar comoving emissivity at 912˚A ( (cid:15) ) versus redshift ( z ). The small blue squaresare taken from the compilations of Khaire & Sri-anand (2015) by integrating QLFs down to 0.01 L ∗ .The magenta open squares are our recalculations oftheir z > z ∼ . z ∼ . quasars reside in the most massive dark matterhalos. Future deep wide field imaging and spec-troscopy are needed to show whether luminousquasars reside in the most biased environment.6.2. Quasar Contribution to Reionization Here, we estimate the quasar contributionto the ionizing photons at z ∼ . α ν = − . α ν = − . ∼ ∗ (Madau et al.1999; Khaire & Srianand 2015).As we have measured Φ = (6 . ± . × − Mpc − mag − at M = − . β = − . ± . 22 at − . 5, the ionization photon produc-tion rate from quasars at z = 6 . z ∼ . z ∼ − . (cid:46) α (cid:46) − . z ∼ z ∼ . β = − . 35 used herewill give a maximum emissivity measurementbecause our single power-law QLF will overesti-mate the number of quasars at the faint-end. Inthis case, the double power-law QLF describedby Eq. (9) with α and characteristic magnitudefixed to the values determined by Jiang et al.(2016) are better suited to estimate quasar co-moving emissivity at z ∼ . 7. If the break mag-nitude evolves following the luminosity evolu-tion and density evolution (LEDE) model (Mc-Greer et al. 2013; Yang et al. 2016), the breakmagnitude at z ∼ . z = 6 (Kulkarni et al. 2018),our best-fit slope is actually the QLF faint-end slope. In this case, the single power-lawwe used here gives the true emissivity measure-ment. But note that, the total quasar comov-ing emissivity is sensitive to the lower bound ofthe integral, especially for the single power-lawcase. Thus, the numbers estimated based onsingle power-law need to be cautionary use. xploring Reionization-Era Quasars (cid:15) = (2 . +4 . − . ) × ergs s − Hz − Mpc − and (cid:15) = (6 . +18 . − . ) × ergs s − Hz − Mpc − , by using the best fit sin-gle power-law QLF ( β = − . 35) and dou-ble power-law QLF ( α = − . β = − . M ∗ = − . z ∼ . N ion = (1 . +4 . − . ) × photons Mpc − s − and ˙ N ion = (6 . +16 . − . ) × photons Mpc − s − for the best-fit single power-law QLF and double power-law QLF, respec-tively.The total required photon rate density tobalance hydrogen recombination was esti-mated by Madau et al. (1999), i.e., ˙ N ion ( z ) =10 . (cid:0) C (cid:1) × (cid:0) z (cid:1) (cid:16) Ω b h . (cid:17) photons Mpc − s − , where we adopt Ω b = 0 . 047 and h = 1 . C is suggested to be ∼ z ∼ − C = 2, the to-tal required photon rate density at z = 6 . . × photons Mpc − s − and quasarsprovide ∼ ∼ C = 3,the required photon rate density changes to4 . × photons Mpc − s − and quasars onlyprovide ∼ ∼ α flux from quasar spec-tra (e.g. Meiksin 2005; Becker & Bolton 2013; z i o n [ p h o t o n s M p c s ] C=2C=3Meiksin+2005Becker & Bolton 2013D'Aloisio+2018Re-calculationsDPL ( = 1.90, = 2.54)Single PL ( = = 2.35) Universe Age (Gyr) Figure 12. The evolution of the comoving produc-tion rate of ionizing photons. The solid and opencircles are production rate of ionizing photons at z ∼ . α flux byBecker & Bolton (2013) and D’Aloisio et al. (2018),respectively. The blue solid line represent the re-quired rate inferred from measurements of the meantransmitted Ly α flux from Meiksin (2005). Thedash-dotted and dashed lines are the required pho-ton rate density to balance hydrogen recombinationby assuming C = 2 and C = 3 Madau et al. (1999),respectively. Davies et al. 2018a; D’Aloisio et al. 2018). Itis related to the total (comoving) emissivity ofthe sources, (cid:15) SL , and can be described by ˙ N S ion = (cid:82) ∞ ν L (cid:15) SL h P ( νν L ) − α s dνν ≈ (cid:15) SL h P α s = A S α MG α S (1 + z ) γ h photons Mpc − s − , where h P is the Planck con-stant, A S = 1 . × , α MG = 1, γ = − . α s = 1 . Wang et al. vide ∼ ∼ M ∗ > − . 5, and –2.35 is the bright-end slope) or the true value (if M ∗ < − . 6, and–2.35 is the faint-end slope). Based on theseresults, it is highly unlikely that high redshiftquasars make a significant contribution to hy-drogen reionization. As we mentioned before,the total quasar comoving emissivity is verysensitive to the lower bound of the integral forthe single power law case. If there is a signif-icant population of extremely faint AGN (i.e. M (cid:38) − SUMMARYIn this paper, we presented the discovery of 16quasars at z (cid:38) . 4, and 5 quasars at z ∼ z (cid:38) . ∼ –25.5 to ∼ –27.5. We havemore than doubled the number of known lumi-nous quasars at z > . z (cid:38) . (cid:38) . < z < . ∼ to a fluxlimit of z P S = 21. We measured the slope ofthe QLF to be β = − . ± . 22 within themagnitude range of − . < M < − . α = − . 9) and characteristic mag-nitude ( M ∗ = − . β = − . ± . 29. If we fix thebright-end slope ( β = − . 05) and characteristicmagnitude ( M ∗ = − . β = − . ± . 22. These mea-surements indicate that the QLF slope does notevolve significantly from z = 6 to z = 6 . − . < M < − . z ∼ . ρ ( < − . . ± . 11) Gpc − mag − using the 1/ V a method. By fitting the spatialdensity at z = 6 and z = 6 . 7, we find a densityevolution parameter of k = − . ± . 18 from z ∼ z ∼ . 7, which corresponds to thequasar number density declining by a factor ofmore than three from z ∼ z ∼ . 7, at a ratesignificantly faster than the average decline ratebetween z ∼ z ∼ z ∼ . e -folding times. The rapid decline ofquasar density within such short time requiresthat SMBHs must grow rapidly from z ∼ . z ∼ z ∼ . z ∼ . (cid:15) = (2 . +4 . − . ) × ergs s − Hz − Mpc − and (cid:15) = (6 . +18 . − . ) × ergs s − Hz − Mpc − , using the best fit singlepower-law QLF ( β = − . 35) and double power-law QLF ( α = − . β = − . z = 6 . N ion = (1 . +4 . − . ) × photons s − Mpc − and ˙ N ion = (6 . +16 . − . ) × photons s − Mpc − xploring Reionization-Era Quasars α flux, our measurements suggest that high red-shift quasars have very low possibility to be thedominant contributor of hydrogen reionization(i.e. only contribute < 7% required ionizing pho-tons).We are collecting high quality optical andNIR spectra with large ground-based telescopes.In the forthcoming publications we will givemeasurements of Gunn-Peterson optical depthsbased on both Ly α and Ly β forests, searchdamping wing signatures to constrain the evolu-tion of neutral fraction at the EoR, provide theBH mass measurements and explore the SMBHgrowth history with the high quality spectrathat we are collecting. We are also observingX-ray emissions from these quasars with Chan-dra , which will give insights the accretion sta-tus of of the earliest SMBHs. This dedicateddataset will be an ideal dataset for investigatingthe cosmic reionization history and SMBH for-mation mechanisms. We are also surveying the[C ii ] emission line from the quasar host galax-ies with ALMA and NOEMA . These observa-tions will provide us excellent opportunities tostudy the SMBH and host galaxy co-evolutionin the EoR. Moreover, this unique quasar sam-ple will provide ideal targets for JWST to in-vestigate the environments and host galaxies ofthese SMBHs.F. Wang, J. Yang, X.-B. Wu and L. Jiangthank the supports by the National Key R&DProgram of China (2016YFA0400703) andthe National Science Foundation of China(11533001, 11721303). J. Yang, X. Fan, M. Yue,J.-T. Schindler and I. D. McGreer acknowledgesupport from the US NSF Grant AST-1515115and NASA ADAP Grant NNX17AF28G. J.R.Findlay, B.W. Lyke, A.D. Myers and E. Haze Nunez acknowledge support from the NationalScience Foundation through REU grant AST-1560461.We acknowledge the use of data obtained atthe Gemini Observatory (NOAO program ID:GN-2018A-C-1), which is operated by the As-sociation of Universities for Research in As-tronomy (AURA) under a cooperative agree-ment with the NSF on behalf of the Gem-ini partnership: the National Science Foun-dation (United States), the National ResearchCouncil (Canada), CONICYT (Chile), the Aus-tralian Research Council (Australia), Ministrioda Ciˆencia e Tecnologia (Brazil) and Ministe-rio de Ciencia, Tecnolog´ıa e Innovaci´on Pro-ductiva (Argentina). We acknowledge the useof Hale telescope, Keck I telescope, LBT tele-scopes, Magellan telescopes, MMT 6.5 m tele-scope and UKIRT telescope. Observations ob-tained with the Hale Telescope at Palomar Ob-servatory were obtained as part of an agreementbetween the National Astronomical Observato-ries, Chinese Academy of Sciences, and the Cal-ifornia Institute of Technology. Observationsreported here were obtained at the MMT Ob-servatory, a joint facility of the University ofArizona and the Smithsonian Institution.We acknowledge the use of BASS, DECaLS,MzLS, PanStarrs, UHS, ULAS and VHS photo-metric data. This publication makes use of dataproducts from the Wide-field Infrared SurveyExplorer, which is a joint project of the Uni-versity of California, Los Angeles, and the JetPropulsion Laboratory/California Institute ofTechnology, and NEOWISE, which is a projectof the Jet Propulsion Laboratory/California In-stitute of Technology. WISE and NEOWISEare funded by the National Aeronautics andSpace Administration.This research uses data obtained through theTelescope Access Program (TAP), which hasbeen funded by the National Astronomical Ob-servatories, Chinese Academy of Sciences, and4 Wang et al. the Special Fund for Astronomy from the Min-istry of Finance in China. Facilities: Gemini (GMOS), Hale (DBSP),KECK (DEIMOS), LBT (MODS), Magellan(FIRE, LDSS3-C), MMT (BinoSpec, MMIRS,Red Channel Spectrograph), UKIRT (WFCam)REFERENCES Allen, J. T., Hewett, P. C., Maddox, N., Richards,G. T., & Belokurov, V. 2011, MNRAS, 410, 860Arnaboldi, M., Neeser, M. J., Parker, L. C., et al.2007, The Messenger, 127, 28Ba˜nados, E., Venemans, B. P., Decarli, R., et al.2016, ApJS, 227, 11Ba˜nados, E., Venemans, B. P., Mazzucchelli, C.,et al. 2018, Nature, 553, 473Ba˜nados, E., Venemans, B. P., Morganson, E., etal. 2015, ApJ, 804, 118Becker, G. D., & Bolton, J. S. 2013, MNRAS, 436,1023Becker, R. H., White, R. L., & Helfand, D. J.1995, ApJ, 450, 559Beckwith, S. V. W., Stiavelli, M., Koekemoer,A. M., et al. 2006, AJ, 132, 1729Best, W. M. J., Liu, M. C., Magnier, E. A., et al.2015, ApJ, 814, 118Bolton, J. S., Haehnelt, M. G., Warren, S. J., etal. 2011, MNRAS, 416, L70Boroson, T. A., & Green, R. F. 1992, ApJS, 80,109Bosman, S. E. I., & Becker, G. D. 2015, MNRAS,452, 1105Byard, P. L., & O’Brien, T. P. 2000, Proc. SPIE,4008, 934Carilli, C. L., & Walter, F. 2013, ARA&A, 51, 105Carilli, C. L., Wang, R., van Hoven, M. B., et al.2007, AJ, 133, 2841Casali, M., Adamson, A., Alves de Oliveira, C., etal. 2007, A&A, 467, 777Chambers, K. C., Magnier, E. A., Metcalfe, N., etal. 2016, arXiv:1612.05560Condon, J. J., Cotton, W. D., Greisen, E. W., etal. 1998, AJ, 115, 1693D’Aloisio, A., McQuinn, M., Davies, F. B., &Furlanetto, S. R. 2018, MNRAS, 473, 560Davies, F. B., Hennawi, J. F., Ba˜nados, E., et al.2018, arXiv:1802.06066Davies, F. B., Hennawi, J. F., Eilers, A.-C., &Luki´c, Z. 2018, ApJ, 855, 106 Decarli, R., Walter, F., Venemans, B. P., et al.2017, Nature, 545, 457Decarli, R., Walter, F., Venemans, B. P., et al.2018, ApJ, 854, 97Dey, A., Schlegel, D. J., Lang, D., et al. 2018,arXiv:1804.08657Dye, S., Lawrence, A., Read, M. A., et al. 2018,MNRAS, 473, 5113Faber, S. M., Phillips, A. C., Kibrick, R. I., et al.2003, Proc. SPIE, 4841, 1657Fabricant, D. G., Fata, R. G., & Epps, H. W.1998, Proc. SPIE, 3355, 232Fan, X., Carilli, C. L., & Keating, B. 2006,ARA&A, 44, 415Fan, X., Narayanan, V. K., Lupton, R. H., et al.2001, AJ, 122, 2833Fan, X., Wang, F., Yang, J. 2018, ApJL,submittedFindlay, J. R., Kobulnicky, H. A., Weger, J. S., etal. 2016, PASP, 128, 115003Gibson, R. R., Jiang, L., Brandt, W. N., et al.2009, ApJ, 692, 758G´orski, K. M., Hivon, E., Banday, A. J., et al.2005, ApJ, 622, 759Greig, B., Mesinger, A., Haiman, Z., & Simcoe,R. A. 2017, MNRAS, 466, 4239Haardt, F., & Madau, P. 2012, ApJ, 746, 125Hewett, P. C., & Foltz, C. B. 2003, AJ, 125, 1784Hook, I. M., Jørgensen, I., Allington-Smith, J. R.,et al. 2004, PASP, 116, 425Hopkins, P. F., Hernquist, L., Cox, T. J., et al.2005, ApJ, 630, 705Irwin, M. J., Lewis, J., Hodgkin, S., et al. 2004,Proc. SPIE, 5493, 411Jiang, L., Fan, X., Ivezi´c, ˇZ., et al. 2007, ApJ, 656,680Jiang, L., McGreer, I. D., Fan, X., et al. 2016,ApJ, 833, 222Kashikawa, N., Ishizaki, Y., Willott, C. J., et al.2015, ApJ, 798, 28Kaurov, A. A., & Gnedin, N. Y. 2015, ApJ, 810,154 xploring Reionization-Era Quasars Kellermann, K. I., Sramek, R., Schmidt, M.,Shaffer, D. B., & Green, R. 1989, AJ, 98, 1195Kirkpatrick, J. D., Cushing, M. C., Gelino, C. R.,et al. 2011, ApJS, 197, 19Khaire, V., & Srianand, R. 2015, MNRAS, 451,L30Knigge, C., Scaringi, S., Goad, M. R., & Cottis,C. E. 2008, MNRAS, 386, 1426Koptelova, E., Hwang, C.-Y., Yu, P.-C., Chen,W.-P., & Guo, J.-K. 2017, Scientific Reports, 7,41617Kulkarni, G., Worseck, G., & Hennawi, J. F. 2018,arXiv:1807.09774Illingworth, G. D., Magee, D., Oesch, P. A., et al.2013, ApJS, 209, 6Lang, D. 2014, AJ, 147, 108Lawrence, A., Warren, S. J., Almaini, O., et al.2007, MNRAS, 379, 1599Lusso, E., Worseck, G., Hennawi, J. F., et al.2015, MNRAS, 449, 4204Madau, P., Haardt, F., & Rees, M. J. 1999, ApJ,514, 648Madau, P., & Haardt, F. 2015, ApJL, 813, L8Madau, P., & Rees, M. J. 2000, ApJL, 542, L69Mainzer, A., Bauer, J., Grav, T., et al. 2011, ApJ,731, 53Matsuoka, Y., Iwasawa, K., Onoue, M., et al.2018, arXiv:1803.01861Matsuoka, Y., Onoue, M., Kashikawa, N., et al.2016, ApJ, 828, 26Matsuoka, Y., Onoue, M., Kashikawa, N., et al.2018, PASJ, 70, S35Mazzucchelli, C., Ba˜nados, E., Venemans, B. P.,et al. 2017, ApJ, 849, 91McQuinn, M., Oh, S. P., & Faucher-Gigu`ere,C.-A. 2011, ApJ, 743, 82Meisner, A. M., Lang, D., & Schlegel, D. J. 2017,AJ, 154, 161McGreer, I. D., Fan, X., Jiang, L., & Cai, Z. 2018,AJ, 155, 131McGreer, I. D., Jiang, L., Fan, X., et al. 2013,ApJ, 768, 105McGreer, I. D., Hall, P. B., Fan, X., et al. 2010,AJ, 140, 370McLeod, B., Fabricant, D., Nystrom, G., et al.2012, PASP, 124, 1318McMahon, R. G., Banerji, M., Gonzalez, E., et al.2013, The Messenger, 154, 35Meiksin, A. 2005, MNRAS, 356, 596Miralda-Escud´e, J. 1998, ApJ, 501, 15 Mortlock, D. J., Warren, S. J., Venemans, B. P.,et al. 2011, Nature, 474, 616Oke, J. B., & Gunn, J. E. 1982, PASP, 94, 586Page, M. J., & Carrera, F. J. 2000, MNRAS, 311,433Pentericci, L., Vanzella, E., Fontana, A., et al.2014, ApJ, 793, 113Planck Collaboration, Aghanim, N., Akrami, Y.,et al. 2018, arXiv:1807.06209Reed, S. L., McMahon, R. G., Martini, P., et al.2017, MNRAS, 468, 4702Reichard, T. A., Richards, G. T., Hall, P. B., etal. 2003, AJ, 126, 2594Richards, G. T., Strauss, M. A., Fan, X., et al.2006, AJ, 131, 2766Risaliti, G., & Elvis, M. 2010, A&A, 516, A89Robertson, B. E., Ellis, R. S., Furlanetto, S. R., &Dunlop, J. S. 2015, ApJL, 802, L19Ross, N. P., Myers, A. D., Sheldon, E. S., et al.2012, ApJS, 199, 3Schindler, J.-T., Fan, X., McGreer, I. D., et al.2018, ApJ, 863, 144Schlegel, D. J., Finkbeiner, D. P., & Davis, M.1998, ApJ, 500, 525Schmidt, G. D., Weymann, R. J., & Foltz, C. B.1989, PASP, 101, 713Selsing, J., Fynbo, J. P. U., Christensen, L., &Krogager, J.-K. 2016, A&A, 585, A87Semelin, B. 2016, MNRAS, 455, 962Shankar, F., Crocce, M., Miralda-Escud´e, J.,Fosalba, P., & Weinberg, D. H. 2010, ApJ, 718,231Shen, Y., Strauss, M. A., Oguri, M., et al. 2007,AJ, 133, 2222Simcoe, R. A., Burgasser, A. J., Bernstein, R. A.,et al. 2008, Proc. SPIE, 7014, 70140UStevenson, K. B., Bean, J. L., Seifahrt, A., et al.2016, ApJ, 817, 141Stocke, J. T., Morris, S. L., Weymann, R. J., &Foltz, C. B. 1992, ApJ, 396, 487Tang, J.-J., Goto, T., Ohyama, Y., et al. 2017,MNRAS, 466, 4568Tsuzuki, Y., Kawara, K., Yoshii, Y., et al. 2006,ApJ, 650, 57Vanden Berk, D. E., Richards, G. T., Bauer, A.,et al. 2001, AJ, 122, 549van Leeuwen, F., Evans, D. W., De Angeli, F., etal. 2017, A&A, 599, A32Venemans, B. P., Ba˜nados, E., Decarli, R., et al.2015, ApJL, 801, L11 Wang et al. Venemans, B. P., Findlay, J. R., Sutherland,W. J., et al. 2013, ApJ, 779, 24Venemans, B. P., Walter, F., Decarli, R., et al.2017, ApJL, 851, L8Vestergaard, M., & Wilkes, B. J. 2001, ApJS, 134,1Volonteri, M., & Rees, M. J. 2006, ApJ, 650, 669Volonteri, M., Silk, J., & Dubus, G. 2015, ApJ,804, 148Wang, R., Carilli, C. L., Beelen, A., et al. 2007,AJ, 134, 617Wang, F., Fan, X., Yang, J., et al. 2017, ApJ, 839,27 (Paper I)Wang, F., Wu, X.-B., Fan, X., et al. 2015, ApJL,807, L9 Wang, F., Wu, X.-B., Fan, X., et al. 2016, ApJ,819, 24Wang, F., et al. 2018, ApJL, submitted (Paper II)Willott, C. J., Delorme, P., Reyl´e, C., et al. 2010,AJ, 139, 906Wright, E. L., Eisenhardt, P. R. M., Mainzer,A. K., et al. 2010, AJ, 140, 1868Wu, X.-B., Wang, F., Fan, X., et al. 2015, Nature,518, 512Wyithe, J. S. B., & Loeb, A. 2003, ApJ, 595, 614Yang, J., Wang, F., Wu, X.-B., et al. 2016, ApJ,829, 33York, D. G., Adelman, J., Anderson, J. E., Jr., etal. 2000, AJ, 120, 1579Yuan, H., Zhang, H., Zhang, Y., et al. 2013,Astronomy and Computing, 3, 65Zou, H., Zhang, T., Zhou, Z., et al. 2017,arXiv:1712.09165 xploring Reionization-Era Quasars Table 5. Observational information of 5 new z ∼ Name Telescope Instrument Exposure (sec) OBS-DATE (UT)DELS J020611.20 − APPENDIX A. DISCOVERY OF FIVE NEW QUASARS AT Z ∼ z ∼ i -dropout method by combing DELS and infraredsurveys mentioned above. The detailed i -dropouts selection was described in Wang et al. (2017) andwill not be repeated here. The only difference is that we used unWISE (Lang 2014; Meisner et al.2017) photometry instead of ALLWISE photometry when selecting z ∼ z ∼ i -band imaging with Wide-field Cameramounted on the Wyoming Infrared Observatory (Findlay et al. 2016). After PS1 DR1 was released,we instead used PS1 i -band as the dropout band. Following Wang et al. (2017), we took spectroscopicobservations for those candidates satisfy i − z > . SN ( i ) < 3. We observed seven z ∼ i -dropouts, we identified five new z ∼ . ≤ z ≤ . 15. However, due to poorer quality of these discovery spectra compared with that of z > . z ∼ y ps -band, we scaled the composite spectra to PS1 y ps -band photometry to measure magnitudes at 1450 ˚A. These quasars have similar luminositieswith that of two quasars reported in Wang et al. (2017) and are fainter than majority quasars foundby SDSS and PS1 quasar surveys (e.g. Fan et al. 2001; Ba˜nados et al. 2016), but brighter than SHEL-LQs quasars (e.g. Matsuoka et al. 2016). Table 6 present the redshift and photometric informationof these five newly discovered z ∼ Wang et al. Table 6. Properties of five newly discovered z ∼ Name Redshift m M i AB z DELS , AB y ps1 , AB J VEGA W VEGAd NIR SurveyDELS J154825.40+005015.5 6.15 ± ± ± ± b ± ± ± ± ± ± ± ± c ± ± ± ± − a ± ± ± ± c ± ± ± ± ± ± ± ± c ± ± ± ± ± ± ± ± b ± ± ± ± a This quasar was discovered by Matsuoka et al. (2018a) independently. b The i -band photometry come from PS1 DR1 photometric catalog. c The i -band photometry were obtained with Wide-field Camera mounted on the Wyoming Infrared Observatory (Findlay et al. 2016). d The WISE W1 magnitudes come from unWISE photometric catalog. f λ ( − e r g s − c m − Å − ) J0206-0255, z=6.08000 8250 8500 8750 9000 9250 9500 Wavelength (Å) 024 J0843+2911, z=6.158000 8250 8500 8750 9000 9250 9500 Wavelength (Å) 012 J1548+0050, z=6.15 Figure 13. Spectra of five newly discovered z ∼ α and N vv