a survey of luminus high-redshift quasars based on sdss and...
TRANSCRIPT
A Survey of Luminus High-Redshift Quasars Based on SDSS and WISE
Xue-Bing Wu (吴学兵)
(KIAA & DoA, Peking University)
East Asia AGN Workshop 2016 (Seoul, 2016/09/22 )
Our Team
Luis Ho Minjin Kim
Telescopes we used 2.4-m telescope, Lijiang, Yunnan, China
MMT(6.5-m) Magellan(6.5-m)
2.16-m telescope, Xinglong/NAOC, China
Bok 2.2-m, Kitt Peak USA
ANU 2.3-m, Australia
Hale 5.1-m
Outline • Problems in finding z~5 quasars • A survey of luminous z~5 quasars with SDSS
and WISE based on 2m class telescopes • Finding z~5.5 quasars • Summary
1. Wang, Wu, Fan, et al., 2016, ApJ, 819, 24 2. Yang, Wang, Wu, et al., 2016, ApJ, 829, 33 3. Yang, Fan, Wu, et al., 2016, to be submitted 4. Wu, Wang, Fan, et al., 2015, Nature, 518,512 5. Wang, Wu, Fan, et al. 2015, ApJL, 807, L9 6. Yi, Wang, Wu, et al., 2014, ApJL, 795, L29
1. Problem in finding z~5 quasars
The most distant quasar at z=7.085 (Mortlock et al. 2011, Nature)
High-z quasars can help us probe the early Universe
Reionization
• The peak of reionization around z~9; from CMB (Planck)
• Reionization ends at z~5-6; from high-z quasars
• The neutral fraction of IGM increases rapidly at z>6.
Fan (2006)
Redshift Distribution of Quasars
An obvious dip at z~5.5
Among 300k known quasars, 200 have z>5 ; 80 have z>5.7; 1 has z>7
Why is it difficult to find z~5.5 quasars? (M star contaminations！)
McGreer et al. 2013； Effective: 4.7< z < 5.1
Fan et al. 2001, 03, 04, 06 Jiang et al. 2008, 09；Effective: z > 5.8
Star
Star QSO
QSO
2 Yang et al.
ever, at z ⇠ 5.5 they are still poorly constrained due tothe lack of a complete quasar sample. McGreer et al.(2013) derived the quasar spatial density at z ⇠ 4, 4.9and 6, and fitted a luminosity-dependent density evolu-tion model to the combined dataset. Their model showsa turnover in number density at z ⇠ 5.5, when assuminga maximum break luminosity M⇤
1450
= -27.0. It meansthat at z � 5 quasars may decline more rapidly thancompared to the rate observed at lower redshift. How-ever, the exact form of the evolution of quasar densityfrom z = 5 to 6 is unclear because of the small size andhigh incompleteness of existing z ⇠ 5.5 quasar sample.The quasar number density at z ⇠ 5.5 is also neededto estimate the contribution of quasars to the ionizingbackground just after the reionization epoch. Willottet al. (2010a) suggested that there was a rapid blackhole mass growth phase after z ⇠ 6. Study of black holegrowth at z ⇠ 4.8 supports that the notion of fast SMBHgrowth at this epoch, corresponding to probably the firstsuch phase for most SMBH (Trakhtenbrot et al. 2011).Studying BH growth properties at z ⇠ 5.5 will fill in themissing link between z ⇠ 5 and 6.To answer the questions posted above, a large, uni-
formly selected sample of quasars at 5.3 < z < 5.7 isneeded. However, so far there has not ever been a com-plete quasar survey at z ⇠ 5.5. As shown in §2, broad-band colors of z ⇠ 5.5 quasars are very similar to thoseof much more numerous M dwarfs, when a small num-ber of passbands are used. Therefore, to avoid the lagernumber of star contaminations, previous quasar selec-tions have always rejected the region of M dwarf locusin r � i/i � z color-color diagram. As a result, most ofpreviously searches for z ⇠ 5.5 quasars have been highlyincomplete. To construct a large uniform z ⇠ 5.5 quasarsample for the study of QLF, SMBH growth and IGMevolution, a more e↵ective selection to separate quasarsfrom M dwarf in this most contaminated region is re-quired.In this paper, we report initial results from a new
search that focuses on the selection of z ⇠ 5.5 quasars.Our new color selection criteria which combine optical,near- and mid-IR colors have yield 17 quasars in the red-shift rage of 5.3 z 5.7 during the pilot observationdescribed here. Our optical/IR color selection techniqueand candidate selection using a combination of existingand new imaging surveys are described in Section 2. Thedetails of our spectroscopy observation and new discov-eries are presented in Section. 3 and Section. 4. InSection 5, we discuss the completeness of our new se-lection. In this section, we also report a test selectionand first discovery using the UKIRT Hemisphere Survey(UHS) photometric data. A summary is given in Section6. In this paper, we adopt a ⇤CDM cosmology with pa-rameters ⌦
⇤
= 0.728, ⌦m = 0.272, ⌦b = 0.0456, and H0
= 70 kms�1Mpc�1 (Komatsu et al. 2009). Photometricdata from the Sloan Digital Sky Survey (SDSS) are in theSDSS photometric system (Lupton et al. 1999), which isalmost identical to the AB system at bright magnitudes;photometric data from IR surveys are in the Vega sys-tem. All SDSS data shown in this paper are correctedfor Galactic extinction.
2. SELECTION OF Z ⇠ 5.5 QUASAR CANDIDATES
2.1. Using Optical and IR colors to Separate Quasarsand M Dwarfs
At z ⇠ 5, most quasars are undetectable in u-band andg-band because of the presence of Lyman limit systems(LLSs), which are optically thick to the continuum ra-diation from the quasar (Fan et al. 1999). Meanwhile,Lyman series absorption systems begin to dominate inthe r-band and Ly ↵ emission moves to the i�band. Ther� i/i� z color-color diagram is often used to select z ⇠
5 quasar candidates in previous studies (Fan et al. 1999;Richards et al. 2002; McGreer et al. 2013). At higher,the i� z color becomes redder and most z > 5.1 quasarsbegin to enter the M dwarf locus in the r� i/i� z color-color diagram, which makes it very di�cult to select z &5.2 quasars only with optical colors, especially at z ⇠ 5.5,where quasars have essentially the same optical colors asM dwarfs (See Figure 1). Previous selections focused onthe region in the right-bottom of the r� i/i�z diagram.IR data are needed to explore the region overlapped withM dwarf locus.
Fig. 1.— Upper: The color track of quasar at z = 5 to 6 (Reddots and line). The step of color track is �z = 0.1. The contoursshow the locus of M dwarfs, from early type to late type. The cyancontours denote M1-M3 dwarfs, the orange contours denote M4-M6 dwarfs and the purple contours denote M7-M9 dwarfs. Clearly,z ⇠ 5.5 quasars are serious contaminated by late type M dwarfs.Bottom: The spectrum of an average of simulated z ⇠ 5.5 quasarscompared with the spectrum of a typical M5 dwarf a. The dashedlines represent normalised SDSS r, i and z bandpasses from leftto right. For the comparison, we scaled both the quasar spectrumand M dwarf spectrum to i = 20.0. In this case, the syntheticSDSS r and z bands magnitudes of quasar are 18.8 and 22.1. ForM5 dwarf, the magnitudes are 18.9 in r band and 21.8 in z band.It is obvious that their r � i and i� z colors are too similar to bedistinguished.ahttp://dwarfarchives.org
Using optical colors can hardly separate z~5.5 quasars and M stars
z~5.5 quasar
z=5~6 Quasar
M1~M3 M4~M6
M7~M9
M5 Star
z=5
z=6
2. A survey of luminous z~5 quasars with SDSS and WISE
The bright end of z ⇠ 5 QLF based on SDSS-WISE selection 5
Fig. 4.— A example of simulated quasar spectrum(blue) at z ⇠ 5 compared with the composite quasar spectra(grey) (Vanden Berk et al.2001; Glikman et al. 2006). AS a comparison, we generate a composite spectra at wavelength bluer than 1780A from SDSS DR12 quasarsat 4.7 < z < 5.4 for absorption and then combine the composite quasar spectras from Vanden Berk et al. (2001) and Glikman et al. (2006)for 1780 - 5700A and � > 5700 A to cover the all wavelength range from SDSS r band to WISE W2 band. The colourful dashed lines showthe transmission curve of SDSS r, i, z and WISE W1, W2 band.
Fig. 5.— The selection function of our survey (zAB < 19.5).The probability is the fraction of simulated quasars which can beselected by our selection criteria in each (M1450,z) bin.
SDSS survey. Therefore within our magnitude limit theSDSS detection can be considered as complete. Our sur-vey added ALLWISE W1 and W2 photometry data intoselection and an new incompleteness was involved dueto the shallower ALLWISE detection. We correct thisincompleteness by using ALLWISE detection complete-ness10 which is considered as a function of flux. The com-pleteness is measured by modelling stellar-like sourcesbased on SDSS DR9 and Two Micron All Sky Survey(2MASS) data (Lake et al. 2013). The red lines in Fig-ure 7 represent the best fit empirical models for W1 andW2 band detection completeness.
4.2. Quasar luminosity function
4.2.1. Quasar sample
From our luminous quasar candidates sample, we havediscovered 64 new quasars with redshifts of 4.4 < z < 5.5.The redshifts of them were measured from Ly↵, Nv,
10 http://wise2.ipac.caltech.edu/docs/release/allwise/
Fig. 6.— Spectroscopic incompleteness of our 110 z ⇠ 5 quasarcandidates. The orange line denotes the spectroscopic incomplete-ness as a function of z band magnitude. The histogram is dividedinto several components filled by di↵erent colors and represents newidentified high redshift quasars (red), low redshift quasars (purple),stars (blue) and unobserved candidates (grey). The low redshiftquasar has been confirmed as a BAL quasar and was selected byour selection due to its redden colours.
O i/Si ii, C ii, Si iv and C iv emission lines (any avail-able) by a eye recognition assistant for quasar spectrasoftware, which is an interactive semi-automated toolkitallowing user to visualize observed spectra and measurethe redshift by fitting the observed spectra to the SDSSquasar template (Vanden Berk et al. 2001), and interac-tively access related spectral line information (ASERA;Yuan et al. 2013). The redshift error measured basedon this method mainly depends on the quality of theobserved spectra and line properties. Due to the low res-olution and strong absorptions at the blueward of Ly↵,the typical redshift error is about 0.05 for Ly↵ basedredshift and less for that based on more emission lines.However, the redshift error could be up to 0.1 for weakemission line quasars and relative low S/N spectra.
We calculate M1450
by fitting a power-law continuum
z~5 quasar spectrum
SDSS WISE
Luminous high-redshift Quasar Survey with SDSS-WISE 5
Fig. 5.— The z�W1 vs. W1�W2 color-color diagram. The pur-ple dashed line represents our selection criteria for quasar candi-dates. The green solid line represents the color-z relation predictedusing quasar composite spectra (Glikman et al. 2006). The solidsquares mark the color tracks for quasars from z = 4.4 to z = 5.4,in steps of �z = 0.2. The grey map denotes SDSS stars, the bluecrosses denote SDSS 4.7 < z < 5.5 quasars, and the red starsdenote our newly discovered quasars.
The purple dashed lines in Figure 4 and Figure 5 de-note our color-color selections (Eq. (5-7)). Apparently,our r � i/i � z selection criteria is much looser thanother studies (the region between purple dashed line andorange dashed line), which will improve the complete-ness of z & 5.1 quasars. Among 274 SDSS and BOSS4.7 z < 5.6 quasars, 22 of them (blue crosses betweenpurple dashed line and orange dashed line in Figure 4)satisfy our r � i/i� z cuts but not the cuts in Richardset al. (2002), and seven (⇠ 32%) of them with z > 5.1.Except the one was not detected by ALLWISE, other sixz > 5.1 quasars are also satisfy our z�W1 / W1�W2selection criteria shown in Figure 5. So we can expectto improve the completeness of selecting z & 5.1 quasarswith our method by combing SDSS and ALLWISE.We started our quasar candidate selection from a cat-
alog of SDSS Data Release 10 (DR10) primary pointsources. Applying the optical magnitudes and colors cuts(Eqs (1)-(5)) using SDSS-III DR10 Query/CasJobs14 re-sults in 457,930 point sources. We then cross-identifiedthese sources with ALLWISE source catalog using a po-sition o↵set within 200; this reduced our candidate listto 80,404 sources with ALLWISE detections. We se-lected our candidates using Eqs (6)-(9) and resulted in1,262 candidates. As we discussed in §2, the ALLWISEdatabase has a high completeness for finding quasarswith z�band magnitudes brighter than 19.5. Limitingcandidates to z 19.5 reduces our candidate list to 420objects. Before conducting spectroscopic observations,we visually inspected images of these 420 candidates, andremoved 231 candidates with suspicious detections, suchas those close to very bright stars or binaries. This listof 189 objects is our primary z ⇠ 5 quasar candidatesample. Removing 78 previously known quasars and oneknown dwarf results in a total of 110 candidates thatrequire spectroscopic followup observations. We have
14 http://skyserver.sdss.org/casjobs/
obtained spectra for 99 of these candidates, and alsore-identified one known quasars (J022112.62�034252.26,See Table 3) that were not published at the time of obser-vations. In addition to our primary sample at z 19.5,we also include candidates fainter than 19.5 as a sup-plementary sample, and observed several of them duringour spectroscopic runs as a test for our ability for findingfainter quasars using SDSS/WISE selection.
3.2. Spectroscopic Observations
Optical spectroscopic observations to identify thesequasar candidates were carried out using a number offacilities: the Lijiang 2.4 m telescope (LJT) and the Xin-glong 2.16m telescope in China, the Kitt Peak 2.3m Boktelescope and the 6.5m MMT telescope in the U.S., aswell as the 2.3m ANU telescope in Australia. We haveobserved 99 candidates from our main sample and 64(64.6%) of them are high-redshift quasars with a redshiftof 4.4 . z . 5.5. We also observed 14 fainter candidatesfrom our test sample and 8 (57.1%) of them are quasarsat 4.7 < z < 5.4. Table 3 lists the observational infor-mation of the 72 new identified quasars.The Lijiang 2.4m telescope is located at Lijiang Ob-
servatory, Yunnan Observatories, Chinese Academy ofSciences (CAS). It is equipped with the Yunnan FaintObject Spectrograph and Camera (YFOSC ), which cantake spectrum followed by photometric images in a veryshort switch time (Zhang et al. 2014). We observed 48candidates by using the YFOSC, with a 2k ⇥ 4k CCD de-tector and three di↵erent grisms based on the brightnessof our candidates. We used Grism 3 (G3), with disper-sion of 172 A/mm and wavelength coverage from 3200to 9200 A, to observe the brightest candidates, Grism 5(G5), with dispersion of 185 A/mm and wavelength cov-erage from 5000 to 9800 A, to observe fainter candidates,and Grism 12 (G12), with dispersion of 900 A/mm andwavelength coverage from 5600 to 9900 A, to observe thefaintest candidates in our sample. We used a 1.008 slit forall three grisms. This slit yields a resolution of R ⇠ 670,R ⇠ 550 and R ⇠ 160 for the G3, G5 and G12 grism,respectively.We observed 35 candidates by using the Red Channel
spectrograph on the MMT 6.5 m telescope. We used the270mm�1 grating centered at 7500 A, providing coveragefrom 5500 A to 9700 A. We used the 1.000 or 1.005 slit basedon seeing, providing resolutions of R ⇠ 640 and R ⇠ 430,respectively.We observed 16 candidates by using Boller and Chivens
Spectrograph (B&C) on Steward Observatory’s 2.3mBok Telescope at Kitt Peak, with the G400 Grating and2.005 slit which gave a resolution of R ⇠ 450 and ⇠ 3400Awavelength coverage.We observed seven candidates by using BAO Faint Ob-
ject Spectrograph and Camera (BFOSC) of the 2.16 moptical telescope at the Xinglong station of the NationalAstronomical Observatories, Chinese Academy of Sci-ences (NAOC). We used the G4 or G10 grating with dis-persion of 198 A/mm and 392 A/mm, respectively. Thewavelength coverage of these two grating are 4000-7800A and 4300-9000 A and spectra resolutions of R ⇠ 330and R ⇠ 110 with a 2.003 slit, respectively.We also used the Wide Field Spectrograph (WiFeS) ,
4 Wang et al.
TABLE 1Optical and WISE Photometry of 723 published z > 4.5 quasars.
Name Redshift Ref r �r i �i z �z Opt W1 �W1 W2 �W2 WISE
J000239.39+255034.80 5.800 16 23.09 0.32 21.51 0.11 18.96 0.05 DR10 16.16 0.06 15.54 0.13 AWJ000552.34�000655.80 5.850 16 24.97 0.51 22.98 0.28 20.41 0.13 DR10 17.30 0.16 17.04 99.0 AWJ000651.61�620803.70 4.510 51 18.29 99.0 99.0 99.0 99.0 99.0 REF 15.20 0.03 14.61 0.04 AWJ000749.17+004119.61 4.780 DR12 21.36 0.06 19.97 0.03 19.84 0.08 DR10 16.85 0.11 16.41 0.26 AWJ000825.77�062604.60 5.929 24 23.91 0.59 23.55 0.63 20.01 0.14 DR10 16.81 0.11 15.68 0.14 AWJ001115.24+144601.80 4.964 DR12 19.48 0.02 18.17 0.02 18.03 0.03 DR10 15.29 0.04 14.69 0.06 AWJ001207.79+094720.23 4.745 DR12 21.40 0.07 19.81 0.04 19.86 0.10 DR10 16.26 0.07 15.80 0.17 AWJ001529.86�004904.30 4.930 32 22.59 0.15 20.99 0.05 20.56 0.12 DR10 99.0 99.0 99.0 99.0 99J001714.68�100055.43 5.011 DR7 21.23 0.06 19.45 0.03 19.55 0.09 DR10 15.94 0.06 15.17 0.09 AWJ002208.00�150539.76 4.528 50 19.39 0.03 18.75 0.02 18.54 0.04 DR10 15.54 0.05 15.11 0.10 AW
Note. — Table 1 is available in its entirety in the electronic edition of the Journal. The first ten rows were shown here for guidance regardingits form and content. The name here are in the format of JHHMMSS.SS+/�DDMMSS.SS. The third column list the reference for each quasar,and the tenth column list the references for the optical magnitudes. Most optical magnitudes are from SDSS DR10 photometric catalog and areGalactic extinction corrected SDSS PSF asinh magnitudes (zAB = zSDSS +0.02 mag). The optical magnitudes come from the reference paper andare in AB system, if this quasar does not have SDSS DR 10 photometry. The last column list the flag of WISE data: AW = ALLWISE catalog,AWR = ALLWISE Reject catalog, AS = ALL-SKY WISE catalog, ASR = ALL-SKY WISE Rejcet catalog, and 99 means no detection in anyWISE catalog. This table only includes quasars that are published before July 2015.
color-color diagram is often used to select z ⇠ 5 quasarcandidates in previous studies (Fan et al. 1999; Richardset al. 2002; McGreer et al. 2013). In Figure 4, we showthe r � i/i � z colors of stars and 274 SDSS and BOSS4.7 z < 5.6 quasars (Schneider et al. 2010; Paris et al.2014), as well as di↵erent z ⇠ 5 quasar selection criteria(Richards et al. 2002; McGreer et al. 2013). The opti-cal color selections shown here (cyan and orange dashedlines) are e↵ective for quasars at z . 5.1, but the se-lection becomes very incomplete for quasars at z & 5.1,when they enter the M star locus on the r � i/i � zcolor-color diagram (Richards et al. 2002; McGreer et al.2013). This is consistent with the color redshift tracks(green solid line) derived from the z ⇠ 5 quasar com-posite spectrum constructed from BOSS quasar spectrausing a median algorithm by us. As we discussed in §2,W1�W2 can be used to reject M type stars e↵ectively;the addition of WISE photometry will allow us to loosenthe typically r�i/i�z cuts to reach higher redshift whilestill be able to reject most late-type star contaminants.Following are our selection criteria:
u > 22.3, z < 20.2 (1)
g > 24.0 or g � r > 1.8 (2)
r � i > 1.0 (3)
i� z < 0.72 (4)
i� z < 0.625⇥ (r � i)� 0.3 (5)
z �W1 > 2.5 (6)
W1�W2 > 0.5 (7)
W1 < 17.0,�W2 < 0.2 (8)
z �W1 > 2.8 or W1�W2 > 0.7, if i� z > 0.4 (9)
where optical magnitudes are Galactic extinction cor-rected SDSS PSF asinh magnitudes and the W1 and W2magnitudes are Vega-based magnitudes. The u-band andg-band cuts are the typical magnitudes limits for dropoutbands (Fan et al. 1999). The z-band magnitude cut isto ensure the accuracy of z-band photometry, since the5� detection of SDSS z-band for point sources at 100 see-ing is about 20.5. The spectral energy distribution of
Fig. 4.— The i � z vs. r � i color-color diagram. The pur-ple dashed line represents our selection criteria for quasar candi-dates. The orange dashed line represents the SDSS z > 4.5 quasarselection criteria (Richards et al. 2002) the cyan dashed line de-notes 4.7 . z . 5.1 quasar selection criteria (McGreer et al. 2013).The green solid line represents the color-z relation predicted usingz ⇠ 5.0 SDSS quasar composite spectra. The solid squares markthe color tracks for quasars from z = 4.4 to z = 5.4, in steps of�z = 0.1. The grey map denotes SDSS stars, the blue crossesdenote SDSS 4.7 < z < 5.5 quasars, and the red stars denote ournewly discovered quasars.
z & 4.5 quasars are mainly dominated by a power-lawspectrum with a slope around ↵⌫ ⇠ �0.5 (Vanden Berket al. 2001), which is flatten than that of M dwarfs. Thisdi↵erence leads to a redder z�W1 color of quasar thanthat of M dwarfs (Eq. (6)). As we discussed in last sec-tion, W1�W2 color can separate quasars and late-typestars very e�ciently; here we require W1�W2> 0.5 (Eq.(7)). We use the magnitude or photometric error cutsof ALLWISE photometric data (Eq. (8)) to ensure theaccuracy of the W1�W2 color. Considering the seriouscontaminations and redder W1�W2 colors for z & 5.2quasars (Figure 3), we also require a more strict z�W1and W1�W2 colors for candidates with i� z > 0.4 (Eq.(9)). Although using Eq. (9) leads to a lower complete-ness of selecting z & 5.2 quasars, it helps reduce starcontaminations significantly.
Select high-z quasar candidates with SDSS-WISE Because WISE detection rate of z>4.5 quasars is 75%, using WISE can help find z~5 quasars; revised selection criteria
(purple dashed lines)
(Wang F., Wu, X.-B., et al. 2016, ApJ.)
Star
Star
QSO
美国WISE卫星 40cm
QSO
Observational Results of 110 candidates (zAB<19.5)
Luminous high-redshift Quasar Survey with SDSS-WISE 9
an integral-field double-beam image-slicing spectrographon the ANU 2.3m Telescope at Siding Spring Observa-tory to observe seven of our quasar candidates. They areobserved using Grating R3000 on WiFeS which gives aresolution of R = 3000 at the wavelength between 5300Aand 9800 A.All spectra taken by 2.4m telescope, 2.16m telescope,
2.3m Bok telescope and MMT telescope were reducedusing standard IRAF routines. The WiFeS data were re-duced with a python based pipeline PyWiFeS (Childresset al. 2014). The flux calibrations of all spectra were ob-tained from standard star observations on the same nightand scaled to SDSS i-band magnitudes for absolute fluxcalibrations.
4. RESULTS
4.1. Discovery of 72 New Quasars at z ⇠5
We have spectroscopically observed 99 candidates withz-band magnitudes brighter than 19.5 and 64 (64.6%) ofthem are quasars with redshifts of 4.4 . z . 5.5 and ab-solute magnitudes of �29.0 . M1450 . �26.4. We alsoobserved 14 fainter candidates selected with the same se-lection criteria, and identified 8 (57.1%) fainter z ⇠ 5quasars with 4.7 < z < 5.4 and absolute magnitudeof �27.1 . M1450 . �26.1. Table 4 lists the redshiftand SDSS, WISE photometry of the 72 newly discoveredquasars and Figure 5 shows the optical spectra of thesenew quasars. The redshifts of these quasars were mea-sured from Ly↵, Nv, O i/Si ii, C ii, Si iv and C iv emis-sion lines (any available). We used an eye recognitionassistant for quasar spectra software (ASERA; Yuan etal. 2013), which is an interactive semi-automated toolkitallowing user to visualize observed spectra and measurethe redshift by fitting the observed spectra to the SDSSquasar template (Vanden Berk et al. 2001), and interac-tively access related spectral line information. The red-shift error measured based on this method mainly de-pends on the quality of the observed spectra and lineproperties. Due to the low resolution and strong absorp-tions blueward of Ly↵, the typical redshift error is about0.05 for Ly↵ based redshift and is smaller for that basedon more than one emission lines. However, the redshifterror could be up to 0.1 for weak emission line quasarsand objects with relatively low S/N spectra. Our quasarsample spans a redshift range of 4.40 z 5.53 and theredshift distribution of these newly identified quasars isshown in the lower panel of Figure 7. As we discussed inSection 2, there is a gap of previously published quasarredshift distribution at 5.2 < z < 5.7 with only 33 pub-lished quasars at this redshift due to low identificatione�ciency; 17 of them identified by the SDSS quasar sur-vey. Among the 72 newly identified quasars, 12 of themare at 5.2 < z < 5.7; this represents an increase ⇠36%of the number of known quasars at this di�cult redshiftrange.In Table 3, columns 3 and 4 list the apparent and ab-
solute AB magnitudes of the continuum at rest-frame1450A. They were calculated by fitting a power-law con-tinuum f⌫ ⇠ ⌫↵⌫ to the spectrum of each quasar. Asmany spectra of our newly discovered quasars do nothave enough continuum coverage to reliably measure theslopes of the continua, we assumed that the slope is theaverage quasar UV continuum slope ↵⌫ = �0.5 (Van-
den Berk et al. 2001). The power-law continuum wasthen normalized to match visually identified continuumwindows that contains minimal contribution from quasaremission lines and from sky OH lines. Figure 7 showsthe absolute magnitude at rest-frame 1450A and redshiftdistribution of our newly discovered 72 quasars and pub-lished SDSS z � 4.5 quasars. The red stars are ournew discovered quasars and blue crosses denote SDSSquasars. The red and blue dashed lines in Figure 7 de-note the mean absolute magnitude of our new quasars(M1450 = �27.2) and SDSS quasars (M1450 = �26.4),respectively. Our newly discovered quasars are system-atically brighter than SDSS quasars and improved thecompleteness of luminous z ⇠ 5 quasars in the SDSSfootprint. More important, 24 of our new discoveredquasars with M1450 < �27.5 and doubled the numberof known quasars (26 z > 4.5 SDSS quasars) at thisbrightness range in the SDSS footprint. In particular, 22of our new quasars are at z > 4.7 with M1450 < �27.5,compared to only 13 previously published SDSS quasarsin this redshift/luminosity range.
Fig. 7.— Upper panel : The M1450 vs. redshift diagram. Thesmall blue crosses denote SDSS z � 4.5 quasars, the red stars de-note our newly discovered quasars. The M1450 are AB magnitudesof quasars at 1450A. Lower panel : The redshift distributions ofnewly identified quasars and known z � 4.5 quasars. Our methodprefers to select z > 5.0 quasars consistent with the predictionfrom the color-z relation shown in Figure 3. Right panel : Thedistribution of M1450. Apparently, our newly discovered quasarsare systematically brighter than SDSS quasars and improved thecompleteness of luminous z ⇠ 5 quasars in the SDSS footprint.
4.2. Notes on Individual Objects
SDSS J013127.34�032100.1 (z=5.19). The radio-loudness defined as the ratio of the rest-frame flux densi-ties in the radio (5GHz) to optical bands (4400A) bands(Kellermann et al. 1989). J0131-0321 is a radio-loudquasar with the radio loudness about 100. J0131-0321is the most luminous z & 5 radio-loud quasar known,with SDSS z-band magnitude 18.01 ± 0.03 and withM1450 = �28.29. The observational properties of thisquasar is discussed in details in a separate paper (Yi etal. 2014).SDSS J022112.62�034252.26 (z=5.02). J0221-0342
was independently discovered by BOSS quasar survey
SDSS-WISE High-z Quasar Candidates
MMT(6.5m)
2.4m
Magellan(6.5m)
76 new quasars (some are very bright)
(Wang Feige et al. 2016, ApJ)
Page 3
Figure 2: The distribution of known quasars and our ned discovered quasars at redshift z > 5. There is anobvious gap at 5.1 < z < 5.7. The number of quasars at 5 < z < 6 is significantly lower than it at lowerredshift range. Up to date, including the new result from MMT run on March 14 & 15 2015, we have found39 new quasars at z > 5, 28 of them at 5.1 < z < 5.7, 13 of them at 5.3 < z < 5.7. Before, there are only 20known quasars at 5.3 < z < 5.7. Our selection can be expected to find more quasars located at the redshiftgap.
Figure 3: Left: The i�z/r� i color-color diagram. The purple dashed line represents our selection criteriafor quasar candidates. The orange and cyan dashed line represent the selection criteria of Fan et al. (2000)and McGreer et al. (2013). The red solid line represents the color-z relation predicted using z ⇥ 5.0 SDSSquasar composite spectra. The solid squares mark the color tracks for quasars from z = 4.4 to z = 5.4, in stepsof �z = 0.1. The grey map denotes SDSS stars, the blue stars denote our new discovered quasars. Right:The z � W1/W1 � W2 color-color diagram. The purple dashed line represents our selection criteria forquasar candidates. The solid squares mark the color tracks for quasars from z = 4.4 to z = 5.6, in steps of�z = 0.2
40 new z>5 quasars
7 z~5 quasars found by 2.16m
35 z~5 quasars found by 2.4m
A rare z=5.18 radio-loud quasar (Yi W.-M., Wang F., Wu, X.-B. et al. 2014, ApJL)
i=18.46 F(1.4GHz)=33 mJy Radio Loudness ~ 100 BH mass ~ 3E9 M¤
Eddington Ratio ~ 3 RLQ template
Magellan/FIRE spectrum
AN ULTRA-LUMINOUS QUASAR AT z = 5.363 WITH A TEN BILLION SOLAR MASS BLACK HOLE AND A METAL-RICH DLA AT z ~ 5
（Wang Feige, Wu, Xue-Bing, Fan Xiaohui et al. 2015, ApJL）
The first low-resolution optical spectrum of this source was obtained with the Lijiang 2.4 m telescope using the YFOSC on 2013 November 25 (UT), then follow-up by Magellan.
Quasar Luminosity Function at z~5 The bright end of z ⇠ 5 QLF based on SDSS-WISE selection 9
S = �2NX
i
ln[�(Mi, zi)]+2
Z Z�(M, z)p(M, z)
dV
dzdMdz ,
(11)Where the first term is a sum over each observed quasarin the sample, and the second term is integrated over thefull range of absolute magnitude and redshift of the sam-ple. (Marshall et al. 1983; Fan et al. 2001a). The p(M,z) is the probability for a quasar which with absolutemagnitude M
1450
and redshift z can be observed by thesurvey. It includes all incompleteness discussed above.The second term means the total number of expectedquasars in the survey with a given luminosity functionand provides the normalization for the likelihood func-tion. The confidence intervals are determined from thelikelihood function by assuming a �2 distribution of �S(= S - Smin). (Lampton et al. 1976).
Firstly, we fix the faint end slope ↵ to -2.03 and getthe best fit luminosity function parameters which arelog�(z = 6)⇤= -8.59±0.33, M⇤
1450
= -26.59 ±0.20 and� = -3.31±0.23. This result is plotted in Figure 10 andshows a good agreement with our binned QLF. Then tosee how the di↵erent value of ↵ will a↵ect our result,we also fixed the faint end slope ↵ to be -1.8 which ismore likely with the results from quasar samples at z �4 citedwillot10, glikman10, masters12, mcgreer13 and -1.5 which is based on the evidence for ↵ at lower redshiftz . 3 (Croom et al. 2009). The value of parameters arelisted in Table 3. When we change the faint end slope↵ from -2.03 to - 1.8 and -1.5, the bright end slope � isflattened but only changes a little. The break magnitudebecomes fainter following the change of �.
After that, we allow the all of four parameters to befree and get a steeper bright end slope of � = -4.04 andvery bright break magnitude of M⇤
1450
= -27.91 with bigerr bars. That is caused by a drop of the number ofquasars at M
1450
< -28.3. In this case, the faint endslope ↵ is also be steeper (↵ = -2.33). Considering that↵ = -2.03 is derived from the combined dataset of S82and DR7 quasar sample which includes more quasars inthe magnitude range of M⇤
1450
> -27 and place a strongconstraint on faint end slope, we adopt the result basedon the fixed ↵=-2.03 as our best fit.
TABLE 3Parameters of MLE fit
↵ � M⇤1450 log�⇤(z=6)
-2.03 -3.31±0.23 -26.59±0.33 -8.59±0.20-1.80 -3.14±0.18 -25.92±0.37 -8.15±0.20-1.50 -3.02±0.15 -25.37±0.35 -7.83±0.17-2.33±0.43 -4.04±2.40 -27.91±1.93 -9.58±1.61
Note. — We fixed the faint slope ↵ to -2.03, -1.8 and -1.5 respec-
tively. ↵ = -2.03 is measured from the combination of SDSS S82 and
DR7 sample in M13. We adopt the result with fixed ↵=-2.03 as our
best fit.
We calculate the confidence regions to explore the de-generacy between bright end slope and break magnitudeand plot it in Figure 11. The regions filled with di↵er-ent colors bound 1 sigma (68.3%), 2 sigma (95.4%) and3 sigma (99.7%) regions, respectively. We generate theprobability countours by calculating the Smin for each(M
1450
, �) point and allowing log�(z = 6)⇤ to be free
Fig. 10.— The double power law fits using maximum likelihoodfitting compared with the binned QLF data form S82 sample andour luminous quasar sample. The result based on fixed ↵ and fourfree parameters are plotted for comparison. We fixed the faint endslope ↵ to -2.03 (purple line), -1.8 (yellow dashed line) and -1.5(cyan dot-dashed line) and do the fits respectively. Then we alsoallow all of four parameters to be free (Green dashed line). Thevalue of parameters are listed in Table 3. When we change thegiant end slope ↵ from -2.03 to-1.8 and -1.5, the bright end slope� be flattened but only changes a little. The break magnitudebecomes fainter following the change of �. When we allow the allof four parameters to be free, we get a steeper bright end slope of� = -4.04. That is caused by the drop of the number of quasars atM1450 < -28.3.
Fig. 11.— Confidence region for � and M⇤1450. The regions
filled with di↵erent shades of grey bound 1 sigma (68.3%), 2 sigma(95.4%) and 3 sigma (99.7%) regions, respectively. For a comparison, we plot our best fit result (red cross) and theresult for ↵ = -1.8 (magenta cross) together with the best fit fromother works. The light blue square denotes the best fit form M13.The wheat star shows the result form Willott et al. (2010b) underfixed ↵ = -1.8 (uncertainties of the fit was not reported). Theyellow square represents best fit from Masters et al.(2012) whichinclude both the faint quasar sample in COSMOS field and brightquasars from Richards et al. (2006). We also plot the points (greysquares) to show the best fits for binned data at z = 2.2 (left) and3.4 (right) bins from BOSS S82 sample Ross et al. (2013).
at each point with the fixed ↵ = -2.03. Figure 10 showsthat our data constrain � to a flatter range of -2.6 < � <-4.4 at 95% confidence than the result from M13 whichshow � < 3.1 at 95% confidence. While the best fit fromM13, � = -4, lies within our 2� region. If more luminousquasars at M
1450
< -28.3 will be discovered, The brightend slope will be measured more accurately.
Yang J., Wang, F., Wu, X.-B. et al. 2016, ApJ, 829, 33
Extended to bright end
12 Wang et al.
TABLE 4Photometry of four z > 5.7 quasars selected by our method.
Name z �z m1450 M1450 mi �mi mz �mz W1 �W1 W2 �W2
J010013.02+280225.8a 6.30 0.01 17.51 -29.26 20.84 0.06 18.33 0.03 14.46 0.03 13.64 0.03J154552.08+602824.0 5.78 0.03 19.26 -27.37 21.27 0.07 19.09 0.05 16.00 0.04 15.16 0.05J232514.24+262847.6 5.72 0.05 19.60 -27.01 21.62 0.17 19.41 0.10 16.19 0.06 15.41 0.10J235632.44�062259.2 6.15 0.02 19.89 -26.85 22.55 0.35 19.78 0.11 16.56 0.10 15.70 0.20
a See Wu et al. (2015) for details.
jecting late type star contaminations. Figure 9 shows theknown z > 5.0 quasars (Table 1 and references therein)and M, L, T dwarfs (Kirkpatrick et al. 2011) on thez�W1/W1�W2 color-color diagram. The cyan cyclesdenote 5.0 < z < 5.5 quasars, the blue triangles rep-resent 5.5 < z < 6.5 quasars and the orange stars arez > 6.5 quasars. Clearly, the optical plus WISE photom-etry selection method can separate quasars from contam-inations at redshift up to z . 6.4, the upper redshift limitof optical surveys. Actually, Carnall et al. (2015) haveused similar method (z�W2/W1�W2) to search z > 5.7quasars in VST ATLAS survey, and have identified twonew z > 6 luminous quasars in ATLAS survey area.
Fig. 9.— The z�W1 vs. W1�W2 color-color diagram. The greensolid line represents the color-z relation predicted using quasarcomposite spectra. The green solid squares mark the color tracksfor quasars from z = 5.0 to z = 7.0, in steps of �z = 0.5. The cyancycles denote 5 z < 5.5 quasars, blue triangles are 5.5 z < 6.5quasars and orange stars are z � 6.5 quasars. The red stars rep-resent our new discovered four z & 5.7 quasars. The black crosses,open triangles and open squares are M, L and T dwarfs from Kirk-patrick et al. (2011).
Motivated by the high detection rate of luminoushigh-redshift quasars as mentioned in section 2 (e.g.M1450 . �26.5 at z . 6 and M1450 . �27.0 atz . 7), we also search for luminous z & 5.7 quasarsby combining the i-dropout technique (e.g., Fan et al.2001b; Jiang et al. 2008; Willott et al. 2007; Banados etal. 2014) and z�W1/W1�W2 colors mentioned above.This method can be used to reject supposititious ob-jects without any photometric follow-up observationsand can reject the majority of contaminations (mainlyL and T dwarfs). We have identified four z & 5.7quasars: J010013.02+280225.8 at z = 6.30 ± 0.01(Wu et al. 2015), J235632.44�062259.2 at z = 6.15 ±
0.02, J154552.08+602824.0 at z = 5.78 ± 0.03 andJ232514.24+262847.6 at z = 5.72 ± 0.1. Table 4 liststhe photometric and redshift information for these fourquasars and Figure 10 shows the optical spectra of thesefour quasars. The spectrum of J0100+2802 was obtainedwith LBT/MODS spectrograph (Pogge et al. 2010) andthe redshift determined from Mg ii emission line in thenear-IR spectrum (Wu et al. 2015). J0100+2802 is themost luminous z > 6 quasar ever known and host a 12billion solar black hole in its center (Wu et al. 2015). Weobserved J1545+6028 with a G5 grating on LJT 2.4mtelescope with a 3600 seconds exposure at April 5th,2014. The spectrum of J2325+2628 was obtained withMMT red-channel at May 9th, 2015. We used 270GPMgrating centered at 8500A and a 1.000 slit. The MMT spec-trum shows J2325+2628 is a weak-line emission quasar(WLQ) and with the current S/N there are no clear emis-sion lines. We observed J2356�0622 with ANU WiFeSwith a 30 minutes exposure at May 15th, 2015. However,the target was marginally detected and hard to make sureit is a high redshift quasar due to the cloudy weather. Afurther one hour exposure was performed with WiFeS atJuly 20th which confirmed J2356�0622 as a quasars atz = 6.15. J2356�0622 is the faintest z > 5.7 quasar inour sample with M1450 = �26.85.
Fig. 10.— The optical spectra of four z & 5.7 quasars discov-ered by using the method mentioned in section 6. J0100+2802 isthe most luminous quasar at z > 5 known to date and has beenreported in Wu et al. (2015).
7. SUMMARY
The SDSS spectroscopic surveys have discovered themajority of quasars known to date. However, it focusesmainly on z < 3.5 quasars and has a lower degree of com-
4 new z>5.7 quasars
Wu et al. 2015, Nature
3. Finding z~5.5 quasars Discovery of 16 new z ⇠ 5.5 quasars 3
Fig. 1.— The color-color selection for 5.3 z 5.7 quasars based on r, i, z,J,H,K W1 and W2 bands photometric data from SDSS,ULAS and ALLWISE. We modified the traditional r� i/i� z color cuts for quasars at z ⇠ 5 to cover quasars at z ⇠ 5.5 and add J , H, Kand W1 & W2 data. Blue dots denote simulated quasars at 5.3 z 5.7 and SDSS z band brighter than 20.8, which is the 5� magnitudelimit of SDSS z band. Grey dots show locus of SDSS DR10 spectroscopically identified M dwarfs. Our new identified z ⇠ 5.5 quasars (redstars), lower redshift quasars (purple stars) from our candidate sample are also plotted. The purple dashed lines represent our selectioncriteria, compared with previous r� i/i� z selection pipelines for z ⇠ 5 quasar (Fan et al. 1999; Richards et al. 2002; McGreer et al. 2013;Wang et al. 2016, in brown, orange, cyan and black dashed line, respectively).
M dwarfs to quasars in the sky than shown in plots, thecontamination would be more serious than shown in over-lapped regions. To improve the e�ciency of selection andreduce the size of candidate sample, we restricted our se-lected in most clear regions (purple dashed lines in Fig.1), which still was able to include most quasars.
2.2. Quasar candidate selection
As discussed above, based on riz color-color diagram,to involve J�W1/W1�W2 and H � K/K�W2 in se-lection would enable us to construct a new pipeline todiscover z ⇠ 5.5 quasars. We then used this pipelineto search quasars in this redshift gap. In optical bands,we still use the r � i/i � z color-color diagram to re-ject earlier type stars and quasars at di↵erent redshiftrange. The published large area covered deep NIR pho-tometric database with Y , J , H and K(Ks) bands areULAS (Lawrence et al. 2007) covering ⇠ 4000 deg2 in thenorthern sky and VHS (McMahon et al. 2013) coveringthe whole southern sky. To expand the available surveyarea, although VHS focus on the southern sky, we alsodo the selection in the overlap area between VHS DR3and SDSS, a ⇠ 800 square degree field. For W1 andW2 bands, we used ALLWISE data. ALLWISE8 pro-gram combined photometric data from the WISE cryo-genic (Wright et al. 2010) and NEOWISE (Mainzer etal. 2011) post-cryogenic survey phases and mapped theentire sky with W1, W2, W3 and W4 (3.4, 4.6, 12, and22 µm) bands. ALLWISE have high detection complete-ness of know quasars over almost all redshifts (Wang etal. 2016). Due to the shallower detections in W3 andW4 bands, we only use W1 and W2 data. Therefore, webuilt a quasar survey for z ⇠ 5.5 quasars based on SDSS,ULAS, VHS and ALLWISE photometric data within a ⇠4800 deg2 field. Because ALLWISE W1& W2 detectionscan only reach the depth around SDSS z band magni-tude ⇠ 20.5, we limit our selection and main candidatesample with z < 20.5.We started the candidate selection from a catalog of
candidates which met our SDSS r � i/i � z color cuts,and selected only pointed sources with relatively clean
8 http://wise2.ipac.caltech.edu/docs/release/allwise/
photometry. We also required drop-outs in u and gbands. We limited our selection area to galactic lati-tude b > 20 or b < �20 due to the quickly increasingstar contaminations at lower galactic latitude. We thencross-matched riz selected objects with ULAS DR10 &VHS DR3 and ALLWISE data using a 200 cross radius.We selected sources which meet our SDSS-ULAS/VHS-ALLWISE color cuts. A J � K color cut was addedhere for further rejection of stars. To ensure the W1 andW2 photometry quality, we limit the signal-to-noise tobe higher than 5 in W1 and 3 in W2. We removed tar-gets with suspicious detections, such as multiple peakedobjects or being a↵ected by bright stars. The e↵ect oncolor selection from the di↵erence between ULAS andVHS photometry systems, especially in K(Ks) band, ismuch smaller than photometric uncertainties. Thus weused the same color cuts for ULAS and VHS detected ob-jects. The selection criteria we used are list as following,also shown in Figure 1.
u > 22.3, g > 23.3, z < 20.5 (1)
g > 24.0 or g � r > 1.8 (2)
r � i > 1.3 (3)
0.5 < i� z < 2.2 (4)
J �W1 > 1.5 (5)
W1�W2 > 0.5 (6)
W1�W2 > �0.5⇥ (J �W1) + 1.4 (7)
W1snr � 5,W2snr � 3 (8)
J �K > 0.8 (9)
H �K > 0.35 and K �W2 > 1.3 (10)
Yang, Fan, Wu, et al., 2016, to be submitted
Optical + IR: r, i, z + J, W1, W2 + H, K, W2
6 Yang et al.
Fig. 3.— The spectra of our 22 new discovered quasars. 21 of them are from our main SDSS-ULAS/VHS-ALLWISE selected z ⇠ 5.5quasar candidate candidate sample. J1016+2541 is from the UHS selected sample (Sec 5.1). The blue vertical lines show the Ly�, Ly↵ andSi iv emission lines. All spectra taken using SSO 2.3m/WiFeS (R ⇠ 3000) are smoothed to R ⇠ 600. All spectra are corrected for Galacticextinction using the Cardelli et al. (1989) Milky Way reddening law and E(B � V) derived from the Schlegel et al. (1998) dust map.
21(+3) new quasars: 3 by 2.4m; 8 by 2.3m; 2 by 2.2m 16 (+2) with 5.3<z<5.7
All zAB<20.5 high-z quasars
4 Yang et al.
3. SPECTROSCOPIC IDENTIFICATIONS
Optical spectroscopy for the identification of quasarcandidates were carried out using several facilities: the6.5m MMT telescope in the U.S., the 2.3m ANU tele-scope in Australia, the Lijiang 2.4 m telescope (LJT) inChina and the Kitt Peak 2.3m Bok telescope. Our ob-servation started in the fall 2014. Up to date, we haveobserved 93 candidates from our main sample and dis-covered 21 new quasars. Three quasars, J0155+0415,J2207�0416 and J2225+0330 were also in our observingplan but had been observed and published earlier as z ⇠5 candidates(Wang et al. 2016). All information of spec-troscopic identifications of these 24 quasars are listed inTable 1.We observed around half part of candidates using the
Red Channel spectrograph (Schmidt et al. 1989) on theMMT 6.5 m telescope. We used the 270lmm�1 gratingcentered at 7500 A (8500 A), providing coverage from ⇠5700 to 9300 A (⇠ 6700 to 10300 A). We used the 1.000or 1.005 slits based on the seeing, providing resolutions ofR ⇠ 640 and R ⇠ 430, respectively.We also used the Wide Field Spectrograph (WiFeS;
Dopita et al. 2007, 2010), an integral-field double-beamimage-slicing spectrograph on the ANU 2.3m Telescopeat Siding Spring Observatory, to observe seven of ourquasar candidates. They were observed using GratingR3000 on WiFeS which gives a resolution of R = 3000 atwavelengths between 5300A and 9800 A.The Lijiang 2.4m telescope is located at Lijiang Obser-
vatory, Yunnan Observatories, Chinese Academy of Sci-ences (CAS). It is equipped with the Yunnan Faint Ob-ject Spectrograph and Camera (YFOSC) which can takespectra followed by photometric images with a very shortswitching time. We observed 48 candidates by using theYFOSC, with a 2k ⇥ 4k CCD detector and three dif-ferent grisms based on the brightness of our candidates.We used Grism 3 (G3) with dispersion of 172 A/mm andwavelength coverage from 3200 to 9200 A to observe thebrightest candidates; Grism 5 (G5), with dispersion of185 A/mm and wavelength coverage from 5000 to 9800A, to observe fainter candidates; and Grism 12 (G12),with dispersion of 900 A/mm and wavelength coveragefrom 5600 to 9900 A to observe the faintest candidates inour sample. We used a 1.008 slit for all three grisms. Thisslit yields a resolution of R ⇠ 670, R ⇠ 550 and R ⇠ 160for the G3, G5 and G12 grisms, respectively.Some candidates observed using the Boller and
Chivens Spectrograph (B&C) on Steward Observatory’s2.3m Bok Telescope at Kitt Peak with the G400 Grat-ing and 2.005 slit which gives a resolution of R ⇠ 450 and⇠ 3400A wavelength coverage.All spectra taken on the 2.4m telescope, 2.3m Bok
telescope, and MMT telescope were reduced by usingstandard IRAF routines. The WiFeS data were reducedwith a python based pipeline PyWiFeS (Childress et al.2014). The flux of all spectra were calibrated using stan-dard stars observed on the same night and then scaled toSDSS i/z bands magnitudes for absolute flux calibration.
4. RESULTS
4.1. New discoveries
From our main candidate sample, we have newly ob-served 93 candidates and discovered 21 quasars. Thereare 15 new quasars in the redshift range of 5.3 z 5.7. Others are lower redshift quasars but all at redshiftz > 5, except one broad absorption line quasar with z =4.56. All spectra are plotted in Figure 3. There are also3 quasars in our target list but already being observedand published as z ⇠ 5 candidates (Wang et al. 2016).2 of them are z ⇠ 5.5 quasars, the other is at z = 5.24.Therefore, we construct a uniform selected z ⇠ 5.5 quasarsample with 17 quasars in the magnitude limit of SDSSz = 20.5.We measure the redshifts by visually matching
observed spectrum to quasar template using aneye-recognition assistant for quasar spectra software(ASERA; Yuan et al. 2013; Wang et al. 2016). Thematching is baed on Ly�, Ly↵, Nv, O i/Si ii, and Si ivemission lines. The typical uncertainty of our redshiftmeasurement is around 0.03 and will be less for higherS/N spectra. We do not include the systematic o↵sets ofLy↵ emission line (e.g., Shen et al.2007), which is typi-cal ⇠ 500 km/s and much smaller than the uncertaintyof matching.We calculate the M
1450
by fitting a power-law contin-uum of each observed spectrum. We assume an averagequasar UV continuum slope of ↵
⌫
= �0.5 (Vanden Berket al. 2001), due to the fact that our spectra do not coverwide enough wavelength range for direct slope measure-ment. We normalize the power-law continuum to matchthe visually identified continuum windows that containminimal contribution from quasar emission lines and skyOH lines. Redshift, M
1450
and photometric informationof our new quasars are listed in Table 2. In figure 2, weplot the redshift distribution of our new quasars, com-pared with all previously known quasars and ULAS/VHSdetected known quasars in the magnitude limit z < 20.5.Our discoveries, including the two which have been pub-lished as z ⇠ 5 quasars, almost double the number ofknown quasars at z ⇠ 5.5 and z band brighter than 20.5.
Fig. 2.— Left : The distribution of known quasars and our newlydiscovered quasars with z band brighter than 20.5 at redshift z >4.9. There is an obvious gap at 5.1 < z < 5.7, especially at z ⇠5.5. Up to date, We use SDSS-UKIDSS-WISE color-color selec-tion have discovered 17 new quasars at 5.3 < z < 5.7 and 7 lowerredshift quasars, including three quasars (dark red) that have beenpublished in our z ⇠ 5 quasar sample (Wang et al. 2016). Ouroptical NIR color selection is e↵ective for finding quasars locatedin this redshift gap. Right : The distribution of our newly discov-ered quasars comparing with all SDSS-ULAS/VHS-WISE detectedknown quasars (z < 20.5).
SDSS-ULAS/VHS-WISE detected
More candidates to be observed!
4. Summary • High-z quasars are rare, but important for probing
the early Universe. • We developed new method to select high-z quasar
candidates with SDSS and WISE, and discovered 76 new quasars at z>4.5, including 40 luminous z>5 quasars and 4 quasars at z>5.7. Most were discovered with 2m class telescopes
• We are expanding the program to z~5.5 quasars. 16 new quasars at 5.3<z<5.7 have been discovered
• The new high-z quasar sample is helpful to study the quasar luminosity function at z~5 & z~5.5
PI: Luis Ho (+ 16 members)
