The near and mid-infrared photometric properties of known redshift z ≥ 5 quasars Free

ABSTRACT

We assemble a catalogue of 488 spectroscopically confirmed very high (z ≥ 5.00) redshift quasars (VHzQ) and report their near- (ZYJHK_s/K) and mid- (WISE W1234) infrared properties. 97 per cent of the VHzQ sample is detected in one or more near-infrared (NIR) band, with lack of coverage rather than lack of depth being the reason for the non-detections. 389 (80 per cent) of the very high redshift quasars are detected at 3.4 μm in the W1 band from the unWISE catalogue and all of the z ≥ 7 quasars are detected in both unWISE W1 and W2. Using archival Wide Field Camera (WFCAM)/United Kingdom Infrared Telescope (UKIRT) and VISTA Infrared Camera (VIRCAM)/Visible and Infrared Survey Telescope for Astronomy (VISTA) data we check for photometric variability that might be expected from super-Eddington accretion. We find 28 of the quasars have sufficient NIR measurements and signal-to-noise ratio to look for variability. Weak variability was detected in multiple bands of Sloan Digital Sky Survey (SDSS) J0959+0227, and very marginally in the Y-band of MMT J0215-0529. Only one quasar, SDSS J0349+0034, shows significant differences between WFCAM and VISTA magnitudes in one band. With supermassive black hole accretion likely to be redshift invariant up to very high redshift, further monitoring of these sources is warranted. All the data, analysis codes and plots used and generated here can be found at: github.com/d80b2t/VHzQ.

1 INTRODUCTION

Very high redshift quasars (VHzQ; defined here to have redshifts z ≥ 5.00) are excellent probes of the early Universe. This includes studies of the Epoch of Reionization for hydrogen (see e.g. Fan, Carilli & Keating 2006a; Mortlock 2016, for reviews), the formation and build-up of supermassive black holes (e.g. Rees 1984; Wyithe & Loeb 2003; Volonteri 2010; Agarwal et al. 2016; Latif, Volonteri & Wise 2018; Valiante et al. 2018; Wise et al. 2019) and early metal enrichment (see e.g. Simcoe et al. 2012; Bosman et al. 2017; Chen et al. 2017)

Supercritical accretion, where |$\dot{M} \gt \dot{M}_{\rm Edd}$|⁠, is a viable mechanism to explain the high, potentially super-Eddington, luminosity and rapid growth of supermassive black holes in the early universe (e.g. Alexander & Natarajan 2014; Madau, Haardt & Dotti 2014; Volonteri, Silk & Dubus 2015; Lupi et al. 2016; Pezzulli, Valiante & Schneider 2016; Pezzulli et al. 2017; Takeo et al. 2018). Thus, one could expect VHzQs to potentially vary in luminosity as they go through phases of supercritical accretion. These signatures of photometric variability should be looked for, noting the rest-frame optical emission is redshifted into the observed near-infrared (NIR) at redshifts z > 5, and relativistic time dilation also stretches the photometric variability of the accretion disc as observed. Fortunately, data are now in place from deep, wide-field NIR instruments and surveys such as the Wide Field Camera (WFCAM) instrument on the United Kingdom Infra-Red Telescope (UKIRT) in the Northern Hemisphere and the VISTA InfraRed CAMera (VIRCAM) on the Visible and Infrared Survey Telescope for Astronomy (VISTA) in the Southern Hemisphere, that are necessary for identifying VHzQs.

Quasars are known to be prodigious emitters of infrared emission, thought to be from the thermal emission of dust grains heated by continuum emission from the accretion disc (e.g. Richards et al. 2006; Hill et al. 2014; Leipski et al. 2014; Hickox et al. 2017). Observations in the mid-infrared (MIR), e.g. ∼3–30 μm allow discrimination between active galactic nucleus (AGN)¹ and passive galaxies due to the 1.6 μm ‘bump’ entering the MIR at z ≈ 0.8–0.9 (e.g. Wright, Eisenhardt & Fazio 1994; Sawicki 2002; Lacy et al. 2004; Stern et al. 2005; Richards et al. 2006; Timlin, Ross et al. 2016) as well as between AGN and star-forming galaxies due to the presence of polycyclic aromatic hydrocarbon at |$\lambda \gt 3\, \mu$|m (e.g. Yan et al. 2007; Tielens 2008).

Jiang et al. (2006) and Jiang et al. (2010) report on the discovery of a quasar without hot-dust emission in a sample of 21 z ≈ 6 quasars. Such apparently hot-dust-free quasars have no counterparts at low redshift. Moreover, those authors demonstrate that the hot-dust abundance in the 21 quasars builds up in tandem with the growth of the central black hole. But understanding how dust first forms and appears in the central engine remains an open question (Wang et al. 2008, 2011).

WISE mapped the sky in four passbands, in bands centered at wavelengths of 3.4, 4.6, 12, and 23 μm. The all sky ‘AllWISE’ catalogue release, contains nearly 750 million detections at high-significance,² of which over 4.5 M AGN candidates have been identified with 90 per cent reliability (Assef et al. 2018). Blain et al. (2013) presented WISE MIR detections of 17 (55 per cent) of the then known 31 quasars at z > 6. However, Blain et al. (2013) was compiled with the WISE ‘All-Sky’ data release, as opposed to the superior AllWISE catalogues. That sample only examined the 31 known z > 6 quasars; our sample has 174 objects with redshift z ≥ 6.00 (with 117 detected in WISE). Bañados et al. (2016) reports WISE W1, W2, W3, and W4 magnitudes for the Panoramic Survey Telescope and Rapid Response System 1 (Pan-STARRS1, PS1; Kaiser et al. 2002, 2010), but with no further investigation into the reddest WISE waveband for the VHzQs.

Critically, we now have available to us new W1 and W2 photometry from the ‘unWISE Source Catalogue’ (Schlafly, Meisner & Green 2019), a WISE-selected catalogue that is based on significantly deeper imaging and has a more extensive modeling of crowded regions than the AllWISE release. For the first time in a catalogue, unWISE takes advantage of the ongoing MIR Near-Earth Object Wide-Field Infrared Survey Explorer Reactivation mission (NEOWISE-R; Mainzer et al. 2014), and achieves depths ∼0.7 mag deeper than AllWISE (in W1/2). This additional depth is a significant advantage in the detection and study of VHzQs in the 3–5 micron regime.

Here, we present for the first time the combined NIR properties (from UKIRT and VIRCAM) and the new MIR unWISE for all the spectroscopically known z ≥ 5.00 quasars. Our motivations are numerous and include: (i) establishing the first complete catalogue of z > 5.00 quasars since the pioneering work from SDSS; (ii) utilizing all the WFCAM and VISTA NIR photometry available for the quasars; (iii) making the first study of NIR and MIR variability of the VHzQ population and (iv) establishing the photometric properties for upcoming surveys and telescopes including the Large Synoptic Survey Telescope (LSST),³ ESA Euclid⁴ and the James Webb Space Telescope (JWST)⁵|$^,$|⁶|$^,$|⁷|$^,$|⁸. We chose redshift z = 5.00 as our lower redshift limit due to a combination of garnishing a large sample, adequately spanning physical properties (e.g. luminosity, age of the Universe) and to highlight the parts of L − z parameter space where z > 5 quasars still wait to be discovered.

This paper is organized as follows. In Section 2, we present the assembled list of the 488 z ≥ 5.00 VHzQs that we have compiled. We then give a high-level overview of the photometric surveys and datasets we use and present the photometry of the VHzQs. In Section 3, we investigate the variability properties of the VHzQs, looking for evidence of supercritical accretion. In Section 4, we calculate how many z > 5 quasars we should expect to find in current data sets. We conclude in Section 5 and present all the necessary details to obtain our data set in the Appendices.

We present all our photometry and magnitudes on the AB system. For NIR, we use the Oke & Gunn (1983) and Fukugita et al. (1996) system which has a zero-point of −48.60. For WISE, the AB-Vega offsets are calculated using the Tokunaga & Vacca (2005) AB system, which has a zero-point of −48.574. So to remain on the −48.60 system, and assuming a flux density F_ν ∝ ν⁰, one needs to subtract 0.026 mag from the zero-points: (ΔW1, ΔW2, and ΔW3)  = (2.673, 3.313, and 5.148). For W4, we use the calibration of Brown, Jarrett & Cluver (2014), ΔW4 = 6.66. These magnitudes are not Galactic extinction corrected.

We use a flat Lambda cold dark matter cosmology with H₀ = 67.7 km s⁻¹ Mpc⁻¹, Ω_M = 0.307, and Ω_Λ = 0.693 Planck Collaboration (2016) to be consistent with Bañados et al. (2016) and all logarithms are to the base 10.

2 METHOD AND DATA

Quasars are generally identified by photometric selection followed by spectroscopic confirmation. Here, we reverse this method obtaining first a list of spectroscopic quasars and then obtain photometric information.

We have compiled a list of 488 quasars with redshifts z ≥ 5.00. We use all the z ≥ 5.00 quasars that have been discovered, spectroscopically confirmed and published as of 2018 December 31 (MJD 58483). We then obtain optical, NIR and MIR photometry for the spectral data set. The NIR data comes from two sources: first, the WFCAM (Casali et al. 2007) on the UKIRT, primarily, but not exclusively, as part of the UKIRT Infrared Deep Sky Survey (UKIDSS; Lawrence et al. 2007). And second, data from the VIRCAM on the VISTA (Dalton et al. 2006; Emerson, McPherson & Sutherland 2006). The MIR, |$\lambda =3{-}30\, \mu$|m wavelength data is from the Wide-Field Infrared Survey Explorer (WISE; Wright et al. 2010; Cutri 2013) mission. For reference, Fig. 1 displays the wavelength and normalized transmission of the filters in question.

2.1 Spectroscopy

We compile the list of all known, spectroscopically confirmed quasars from the literature. This list was complied from a range of surveys and papers. Specifically, we use data from: Bañados et al. (2014), Bañados et al. (2016), Bañados et al. (2018a), Bañados et al. (2018b), Becker, Bolton & Lidz (2015), Calura et al. (2014), Carilli et al. (2007), Carilli et al. (2010), Carnall et al. (2015), Cool et al. (2006), Douglas et al. (2007), De Rosa et al. (2011), Fan et al. (2000), Fan et al. (2001), Fan et al. (2003), Fan et al. (2004), Fan et al. (2006b), Fan et al. (2019b), Goto (2006), Ikeda et al. (2017), Jiang et al. (2008), Jiang et al. (2009), Jiang et al. (2015), Jiang et al. (2016), Kashikawa et al. (2015), Khorunzhev et al. (2017), Koptelova et al. (2017), Kozłowski et al. (2019), Kim et al. (2015), Kim et al. (2019), Kurk et al. (2007), Kurk et al. (2009), Leipski et al. (2014), Mahabal et al. (2005), Matsuoka et al. (2016), Matsuoka et al. (2018a), Matsuoka et al. (2018b), Mazzucchelli et al. (2017), Morganson et al. (2012), Mortlock et al. (2009), Mortlock et al. (2011), McGreer et al. (2006), McGreer et al. (2013), Reed et al. (2015), Reed et al. (2017), Saxena et al. (2018), Schneider et al. (2007), Stern et al. (2007), Tang et al. (2017), van Breugel et al. (1999), Venemans et al. (2007), Venemans et al. (2012), Venemans et al. (2013), Venemans et al. (2015b), Venemans et al. (2015a), Venemans et al. (2016), Wang et al. (2016), Wang et al. (2017), Wang et al. (2018a), Wang et al. (2019a), Willott et al. (2007), Willott et al. (2009), Willott et al. (2010), Willott, Omont & Bergeron (2013), Willott, Bergeron & Omont (2015), Wu et al. (2015) Yang et al. (2018a), Yang et al. (2019) and Zeimann et al. (2011).

We updated our list of objects between the submitted paper and accepted due to feedback from other scientists. We discuss these changes in detail in Appendix  C.

Most of these objects are easily identified by their broad Ly α emission line, N v emission and characteristic shape blueward of 1215 Å in the rest-frame. As we shall see, some of the recently discovered objects are close to the galaxy luminosity function characteristic luminosity M*, and some have relatively weak or maybe even completely absorbed Lyα (e.g. figs 7 and 10 in Bañados et al. 2016). We leave aside detailed investigation and discussion into spectral features and line strengths, and take as given the published spectra and redshift identifications.

The breakdown of how many VHzQ each survey reports is given in Table 1. The Sloan Digital Sky Survey (SDSS) and the Pan-STARRS1 (PS1; PSO in Table 1) survey alone identified over half (55.5 per cent) of the VHzQ population. Data from the Hyper Suprime-Cam (HSC) on the Subaru telescope is responsible for 13.1 per cent of our dataset (HSC+SHELLQs in Table 1). The combination of surveys is also vital for identifying VHzQs. The UKIDSS Large Area Survey (ULAS) on its own, or in combination with other surveys is responsible for 6.3 per cent of the sample (SUV+ULAS) including the highest-z object. Where more than one survey is used for the high-redshift identification (e.g. via shorter-band veto and longer wavelength detection) we follow the discovery paper naming convention.

The redshifts for the VHzQs generally come from the measurement of broad ultraviolet (UV)/optical emission lines. There are far infrared emission lines e.g. C ii 158  μm available for several objects, but at the level of our current analysis broadline redshifts are sufficient.

The number of objects at or above various redshifts, along with the corresponding age of the Universe is given in Table 2.

The N(z) redshift histogram is given for the sample in Fig. 2. We split the contribution up by survey. For clarity, we show the individual surveys of SDSS, PS1, HSC, the ULAS detection, and tally the remaining surveys together (‘various’).

2.2 NIR photometry

The NIR data in this paper comes from the Wide Field Astronomy Unit’s (WFAU) Science Archives for UKIRT-WFCAM, the WFCAM Science Archive (WSA; Hambly et al. 2008) and VISTA-VIRCAM, the VISTA Science Archive (VSA; Cross et al. 2012). These archives were developed for the VISTA Data Flow System (VDFS; Emerson et al. 2004).

We access both the WSA and the VSA and include all non-proprietary WFCAM data, which covers all public surveys and PI projects from Semester 05A (2005 May 01) to 2017 January 01, and all non-proprietary VISTA data, which covers all public surveys and PI projects from science verification on 2009 October 15 to 2017 January 01.

Here, we are not just querying the WSA or VSA data tables. We are taking a list of objects (positions) are performing matched aperture (‘forced’) photometry on the NIR imaging data. As such, we generate a set of tables that are different in subtle ways to the regular ‘Detection’ tables. The two most important tables for our needs are the [w/v]serv1000MapRemeasurement and [w/v]serv1000MapRemeasAver.

We produce and provide two new databases with all the necessary quantities and measurements to fully reproduce our tables, figures and results herein. Moreover, these databases contain considerably more information than we report here. Full documentation can be found at the WSA Schema Browser and the VSA Schema Browser.

Fig. 3 shows the areal coverage of the ULAS, UHS, VHS, and VIKING. UHS DR1 is 12 600 deg² (J-band); ULAS DR1 3700 deg² VHS DR6 16 000 deg² and VIKING DR5 is 1300 deg². The overlap between UHS DR1 and VHS DR6 is 28 deg. These four surveys together cover 33 000 deg². VHS, ULAS, and UHS do not observe in the Z-band.

2.2.1 Averaging matched photometry

The data was processed using a matched-aperture photometry method where flux is measured at the spectroscopic position of the quasar, without necessarily knowing if there is a formal detection in the NIR photometry beforehand. The matched-aperture pipeline is discussed in Cross et al. (2013) and with fuller details to appear in a forthcoming paper (Cross et al., in preparation).

We query the WSA and VSA performing matched-aperture photometry at the positions of our 488 VHzQs. This database is world-readable and we give the full recipe and relevant SQL queries for accessing both databases in Appendix  A as well as online.

The photometry in a single epoch image often has a low signal-to-noise (SNR) ratio. The advantage of matched aperture photometry on quasars is that co-adding is relatively simple if each epoch is taken in the same aperture and the aperture photometry has been corrected to total. Indeed, the standard aperture corrections work well for point sources. Coadding using the matched-aperture photometry can give better results for the photometry, where the individual epochs are taken from multiple projects with different field-centres and orientations and point-spread functions since the individual epoch scattered light, pixel distortion and aperture corrections can be applied with the correct weighting.

We calculate a set of averaged catalogues, for each pointing and filter, based on the requirements in RequiredMapAverages, in these cases over time spans of 7, 14, 30, 91, 183, 365, and 730 d, over 10 epochs and over all epochs. The averaging process starts at the first epoch and works onwards from there. Again, we present these measurements in the new SQL tables.

We detect 367 unique quasars in the WFCAM WSA database, 237 quasars are detected in the VISTA VSA database with 131 objects in common with both WFCAM and VISTA data. We give the necessary SQL queries syntax at d80b2t/VHzQ.

2.3 MIR data

The MIR data for this study comes from the Wide-field Infrared Survey Explorer (WISE) mission, and we utlize data from the WISE cryogenic, the NEOWISE (Mainzer et al. 2011) post-cryogenic and NEOWISE Reactivation Mission (NEOWISE-R; Mainzer et al. 2014) survey phases.

We use data from the beginning of the WISE mission (2010 January; Wright et al. 2010) through the fifth-year of NEOWISE-R operations (2018 December; Mainzer et al. 2011). There are several major data releases and catalogues based on the WISE mission. Here, we use two: the WISE AllWISE Data release and the recently released ‘unWISE Catalogue’ (Schlafly et al. 2019). The AllWISE program combines the W1 and W2 Single-exposure data from all the WISE survey phases (4-Band Cryo, 3-Band Cryo and Post-Cryo; 2010 January 07 thru 2011 February 01) survey phases, and the W3 and W4 from the 4-Band Cryo phase. The unWISE effort⁹ is the unblurred coadds of the WISE imaging using the AllWISE and NEOWISE-R stacked data (Lang 2014; Meisner, Lang & Schlegel 2018a,b).

For the two shorter WISE bands, λ_eff3.37 μm W1 and λ_eff4.62 μm W2 we generally report the deeper, unblurred unWISE coadd data. For the two longer WISE bands, λ_eff12.1 μm and λ_eff22.8 μm we use the AllWISE Data Release. Most objects in the AllWISE Source Catalogue are unresolved, so the best photometric measurements to use are the deep detection profile-fit photometry measures, wxmpro, wxsigmpro, and wxsnr. The unWISE Catalogue absolute photometric calibration derives from the photometric calibration of the unWISE coadds (Meisner, Lang & Schlegel 2017), which is tied to the original WISE zero points through aperture fluxes in a 27.5arcsec radius.

Previous works (e.g. Krawczyk et al. 2013; Ross et al. 2015; Bilicki et al. 2016) found that cross-matches performed with a radius of 2–3 arcsec between the user catalogue and WISE was a good compromise between completeness and contamination (see e.g. fig. 4 of Krawczyk et al. 2013). We thus use a cross-match radius of 2.75 arcsec. When querying the AllWISE catalogues, ‘Cone Search Radius’ in the AllWISE table search was set to 2.75 arcsec for the Spatial Constraints. The ‘One to One Match’ was ‘not’ checked; although possible, we consider it highly unlikely there would be more than one MIR source contributing to the flux of a single UV/optically bright rest-frame quasar. As a check, we utilize data from the Spitzer Space Telescope in IRAC Channel 2 4.5 μm, IRAC Channel4 8 μm, and MIPS 24 μm, and number counts and luminosity density measurements from Driver et al. (2016, see their figs 8.11, 8.12, and 8.13). The redshift distribution of the sampled 8–24  μm foreground objects is narrow enough and their K-corrections advantageous enough, that the objects counts sharply turnover for AB(W3, W4) ≳ 20 mag, and the normalized differential counts – especially in e.g. WISE W4 – have most of their extragalactic background light energy emerging at AB∼18–20 mag. As a consequence, the relative number or probability of contaminating objects at AB> 20 mag inside our apertures should be small.

Knowing we have secure detections in the NIR bands, and wanting to boost the number of WISE W3/W4 detections, we allow ourselves to be less conservative in querying the AllWISE catalogues and also query the AllWISE Reject Table. However, with the exception of one object (SHELLQs J1208-0200), the AllWISE Reject Table does not contain any further W3/W4 detection information.

For the our MIR variability investigations, we use the NEOWISE-R 2019 Data Release. NEOWISE 2019 makes available the 3.4 and 4.6 μm (W1 and W2) single-exposure images and extracted source information that was acquired up until 2018 December 13 (MJD 58465) including the fifth year of survey operations of NEOWISE. These fifth year NEOWISE data products are concatenated with those from the first 4 yr into a single archive from 2013 December 13 (MJD 56639). The WISE scan pattern leads to coverage of the full-sky approximately once every 6 months (a‘sky pass’), but the satellite was placed in hibernation in 2011 February and then reactivated in 2013 October. Hence, our light curves have a cadence of 6 months with a 32 month sampling gap.

3 RESULTS

Having collated the sample of 488 VHzQs, and obtained their near- and MIR photometry we report here the various photometric properties of the quasars. Table 3 represents the culmination of this effort, is the main data product of this paper and we now examine the assembly of its contents in more detail.

First, we will concentrate on detection rates in the infrared, go on to report on the colour-redshift and colour–colour properties of our sample. We also investigate the photometric variability.

3.1 Detection rates in the NIR

Table 4 gives the detection rates for the VHzQs in the NIR YJHK/K_s-bands. The first thing to note is that the coverage of the NIR surveys for example from the UKIDSS LAS and VISTA VHS, does not overlap the full area for where the VHzQs are detected.

There are 15 objects that have no NIR detections. Three of these (PSO J053.9605–15.7956, PSO J056.7168–16.4769, and DELS J0411–0907) have been observed (by VHS) but are out of our queried time range (for which the data is publicly available). Six objects are at declination δ < 0 deg and have not been observed (or at least the data is not in the VSA archive yet). Six objects are at a declination δ ≥ +60 deg are too far north for UKIRT and cannot be observed.

We note that although we are often working close to the magnitude limits of the surveys, Eddington bias does not a significant effect on our measurements, since we are not discovering new sources, but measuring properties of already known quasars. Indeed in the NIR WFCAM/VISTA data, using forced photometry reduces the Eddington bias to negligible levels since a flux will be measured regardless of whether the source is just bright enough (or too faint) to be detected above the background noise.

3.1.1 Comparing WFCAM and VISTA

There are 131 quasars observed in the overlapping area between WFCAM and VISTA. We used the VegaToAB values to put these objects on the same AB system, and for each object compared the two measurements. First, we calculated the weighted average (calibrated flux) in each filter of both and calculated the ratio and difference between each measurement and the average. Then for each filter we calculated the weighted average of the differences (in mag) for each instrument to see if there were significant offsets. The results are given in Table 5. The only filter with a significant offset is the Y-band. All of the VISTA averages are negative and all of the WFCAM ones are positive, except for J where the difference is negligable. The K_s versus K band comparison may be affected the different shapes of the filters, with K being significantly wider than K_s, able to detect light at longer wavelengths.

We have also checked for quasars with large differences in magnitude between the WSA and VSA. Objects were selected where the average flux was >0. and >5 average flux error. In order to account for large errors in either the WFCAM or VISTA photometry, objects with δmag < 2 × |$\sigma _{\delta \rm mag}$| in either WSA or VSA were removed. This selects one quasar with large, δmag > 0.2 mag differences between the WSA and VSA. This is SDSS J0349+0034 which has a WSA K-band magnitude of 21.03 ± 0.35 mag AB (observed at MJD = 54062.4), while for the VSA K_s-band this is 20.20 ± 0.14 mag AB (observed 3 times between MJD = 55537 and 55542), meaning that the WFCAM and VISTA observations are ∼4 yr apart.

3.2 Detection rates in the MIR

Unlike the NIR coverage, the WISE satellite and mission performed an all-sky survey, so the location of every VHzQ in our data set is covered. However, the depth of the WISE survey depends heavily on sky location, with locations near the Ecliptic Poles having the highest number of exposures.

Before reporting on the detection rates, we investigate this effect using the AllWISE Source Catalogue. Fig. 4 shows the WISE AllWISE magnitude versus SNR, colour-coded by wxcov the mean coverage depth, in each corresponding band. The w123cov values are the mean pixel coverage in W1/2/3 from the W1/2/3 Atlas Tile Coverage Map within an aperture of circular area with a radius of 8.25 arcsec centered on the position of this source. For w4cov this radius is 16.5 arcsec (the AllWISE Source Catalogue column descriptions has further details). The wxcov value takes into account e.g. individual pixels in the measurement area that may be masked or otherwise unusable (reducing the effective pixel count and thus the mean coverage value) as well as pixels that are affected by distortions across the across the focal plane in single-exposure images (where this distortion is corrected when coadding to generate the Atlas Images).

In the two shorter bands W1/2 we see the clear and expected trend for brighter objects to have larger SNR, and also for the higher SNR for objects with more exposures at a given magnitude. The behaviour for the W3/4 bands is different, with two populations clearly evident in W3 and although a bit more mixed, also in W4. With the suggested split at SNR>2, and no obvious R.A./Dec. dependence seen, this behaviour is explained by the fact that there are non-detections in W3/4 for objects (with high W1/2 SNR) that are reported in the AllWISE catalogue.

Table 6 gives the detection rates for the VHzQs in the MIR WISE W1–4 bands. The unWISE depths are impressive with nearly 80 per cent of all the VHzQs being detected in unWISE W1. A total of 12 out of 14 (86 per cent) in unWISE W1 (9 our of 14; 64 per cent in unWISE W2) of the z ≥ 6.78 quasars are detected and moreover all of the z ≥ 7 quasars are detected in both unWISE W1 and W2. This bodes very well for future, small mirror infrared missions (e.g. Ross et al. 2019).

Secrest et al. (2015) present an all-sky sample of ≈1.35 million AGN meeting a two-colour, (W1–W2) versus (W2–W3), photometric selection as applied to sources from AllWISE. This colour selection is motivated by Mateos et al. (2012). We investigate if this catalogue, which is tuned for, and noted to be highly complete at, lower, z ≲ 2 redshifts, contain the VHzQs. We find only five VHzQs from our sample match the Secrest catalogue, but given a W3 detection is required, this is not surprising. Further, in addition to the relative lack of sensitivity in W3, the MIR colors are moving out of the Mateos et al. (2012) colour criterion at high redshift. This was shown using an AGN spectral energy distribution where AGNs above a redshift of about ∼2 fall out of the colour criterion, see e.g. fig. 5 of Mateos et al. (2012) and fig. 9 of Secrest et al. (2015). Assef et al. (2018) released two large catalogues of AGN candidates identified across 30 000 deg² of extragalactic sky from the WISE AllWISE Data Release. The ‘R90’ catalogue, contains 4.5 M AGN candidates at 90 per cent reliability (≈150 AGN candidates per deg²) while the ‘C75’ catalogue consists of 20.9 M AGN candidates at 75 per cent completeness (≈700 AGN candidates per deg²). Crossmatching our catalogue of 488 VHzQs with these catalogues, produces 49 matches with the R90 sample and 129 matches with the C75 sample. Both catalogues unsurprisingly match to the ultraluminous quasar SDSS J0100+2802 (Wu et al. 2015) while the C75, but not the R90 catalogue matches to the first redshift z > 7 quasar ULAS J1120+0641 (Mortlock et al. 2011). Neither the R90 or C75 catalogue matches to the current highest redshift object J1342+0928 (Bañados et al. 2018a).

3.3 MIR colours

Due to the depth and all-sky coverage of the WISE [and NEOWISE-(R)] mission, several investigations have looked at how WISE detects AGN (e.g Assef et al. 2012; Stern et al. 2012; Secrest et al. 2015; LaMassa et al. 2017; Assef et al. 2018; Glikman 2018; Hviding et al. 2018; LaMassa et al. 2019)

Fig. 5 shows the (W1–W2) versus (W2–W3) colour–colour diagram for the 330 VHzQs that have reported w3mpro values (blue to red coloured points), though we note that only 51 (18 per cent) of these objects have w3snr > 3.0. Approximately 400 000 quasars at redshift z ≲ 3.0 from the SDSS Quasar Catalog using the Fourteenth data release (DR14Q; Pâris et al. 2018) are also shown (grey colour scale).

The VHzQ with the bluest colour in (W1–W2) is SDSS J0927+2001, with (W1–W2)  = –0.283 (i.e. 0.328 in Vega), and the WISE W1W2W3 colour image is shown in Fig. 6(a). The reddest object in (W1–W2) is PSO J135.3860+16.2518 with (W1–W2)  = 0.721 (1.33 Vega).

The set of VHzQ detected in WISE W3 and W4 contains 59 objects that are detected in the broad W3 filter and 10 objects that are formally detected in W4 (w4snr≥3.0). Three of the W3/4 objects are presented in Fig. 6. UHS J0439+1634 (Fig. 6(b)) with W4 = 7.165 ± 0.12 was discovered by Fan et al. (2019b) and is a strongly lensed quasar at z =6.51. This high luminosity is mostly not intrinsic, but is boosted by an intervening redshift z ∼0.7 galaxy. Other high-redshift quasars that are bright W3/4 objects e.g. SDWISE J0306+1853 (Fig. 6 (c)) may well also so be lensed (e.g. Glikman et al. 2018; Pacucci & Loeb 2019; Fan et al. 2019b), but high-resolution follow-up is needed to confirm. SDSS J0100+2802 (redshift z = 6.33; Fig. 6 (d)) does not have a formal W4 detection (w4snr = 2.4), but as reported by Wu et al. (2015), has a detection in the 2MASS J, H, and K_s bands and is also the bluest object in both (W1–W3) and (W2–W3).

3.4 Variability

VHzQs, if accreting at, or above the Eddington limit, might well have changing mass accretion rates, i.e. |$\ddot{m}_{\rm accr}$|⁠. A consequence of this would be that these quasars exhibit signs of variability, most likely showing up in their UV/optical rest-frame emission. We look for evidence of this variability signature in the NIR and MIR light-curves of the VHzQs.

Quasars are known to have dramatically changing Balmer lines, especially Hβ (e.g. LaMassa et al. 2015; MacLeod et al. 2016; Ruan et al. 2016; Runco et al. 2016; Runnoe et al. 2016; Gezari et al. 2017; Assef et al. 2018; Ross et al. 2018; Stern et al. 2018; Yang et al. 2018b; Graham et al. 2020; MacLeod et al. 2019). As a guide, we note that Hα is redshifted to 3.94 μm (i.e. W1) at z = 5.00 and 5.57 μm, which is between the W2 and W3-bands at z = 7.50. Hβ is redshifted to 2.92 μm, which is the blue edge of W1, at z = 5.00 and 4.13 μm at z = 7.50. Less well understood is the temporal behaviour of the metal lines, in particular C iv and Mg ii . C iv enters the Y-band at redshift z = 5.32 and exits at z = 5.99, and enters the J-band at redshift z = 6.55 and exits at z = 7.57. Mg ii enters the H-band at redshift z = 4.33 and exits at z = 5.37 and enters the K-band at redshift z = 6.25 and exits at 7.50.

One key point to note is that the NEOWISE-AllWISE match radius is 3.00 arcsec and the AllWISE (and 2MASS) source information included in the NEOWISE Source Database are associations not identifications.

Fig. 7 gives the number of NEOWISE-R epochs and detections for each VHzQ. There are 74 quasars with over 100 NEOWISE-R epochs, nine with over 200 epochs and four with over 300. Fig. 8 presents three examples of the brighter MIR objects with well-sampled MIR lightcurves and associated colour changes; SDSS J0100+2802 (Fig. 8 top), SDWISE J1752+5036 (middle), SDSS J1621+5155 (bottom). For the light curves, we show weighted means for every 6 month NEOWISE-R scan. Due to the WISE scanning pattern, objects close to the North Ecliptic Pole are well sampled, and indeed this is the area for further investigation time-domain investigation with upcoming large infrared space telescopes (Jansen & Windhorst 2018; White et al. 2019).

Using the extended datasets described in Section 2.2, we select objects with at least eight measurements observed with intervals of at least 30 d, in at least one filter. The average calibrated flux over all epochs must be >0 (i.e. aperJky3>0) with the SNR is required to be ≥8 (i.e. aperJky3/aperJky3Err≥8). With these criteria, there are 18 WFCAM and 12 VISTA objects with 2 objects (SDSSJ0221-0342B |$\&$| VIK J1148+0056) in common to both, and this sample of 28 is our starting point for variability investigations.

We present four objects that are particularly well-sampled in the NIR. These are: MMT J0215–0529 (z = 5.13; Fig. 9), CFHQS J0216–0455 (z = 6.01; Fig. 10), SHELLQs J0220–0432 (z = 5.90; Fig. 11) and SDSS J0959+0227 (z = 5.07; Fig. 12). We incorporate the MIR NEOWISE-R light curve data where available.

We note that sometimes the average flux is negative and so a default magnitude is reported. This is apparent for the average H and K_s flux for CFHQS J0216–0455 at MJD = 56900. Interestingly this is evident in both the H and K_s filters, but with this quasar not much brighter than the detection limit, one must be very careful with any (over) interpretation.

Even with well-sampled data across the 3000–4000 d observed time-scales that the WFCAM and VISTA surveys span, the (1 + z) time-dilation dramatically affects the sampled rest-frame time-scales sampled, which are ∼300–700 d. Indeed, when sharp changes in a accretion rate are expected on the system’s dynamical time-scale of several kiloyears (e.g. Regan et al. 2019), then seeing any variability signature is not expected. However, noting that (i) accretion on to supermassive black holes should be generally time invariant, and (ii) that in lower redshift quasars, which also have massive >10⁸ M_⊙ black holes, dramatic changes in both continuum and line emission are seen on much shorter time-scales, continued monitoring of these objects is definitely warranted.

We finally focus on SDSS J0959+0227 which is presented in Wang et al. (2016) but is first reported in Civano et al. (2011) as CID-2220 and a high-redshift, z > 3 AGN, in the Chandra-COSMOS field. The spectrum from this object is presented in Ikeda et al. (2012) and shows SDSS J0959+0227 having a narrow Ly α line, and would likely be a Lyman-α emitting galaxy had it not been an X-ray source, The X-ray luminosity is ≈3 × 10⁴⁴ erg s⁻¹ in the 2–10 keV rest-frame, so it is an AGN. This object is clearly not a regular broadline ‘Type 1’ AGN. Noting that the redshift of z = 5.07 for SDSS J0959+0227, the Y, J, H, Ks/K-bands correspond to rest-wavelengths of ∼1690, 2055, 2690, 3515/3625 Å i.e. the rest-frame UV/very blue, so, so the regular blue QSO continuum could be emerging. Thus, with the slightest hint of variability, this could potentially be a high-z AGN transitioning from a narrow-line ‘Type 2’ object to a broadline Type 1 quasar.

4 CONCLUSIONS

In this study, we have, for the first time, compiled the list of all z > 5 spectroscopically confirmed quasars. We have assembled the NIR (Z, Y, J, H, K/K_s) and MIR (WISE W1/2/3/4) photometry for these objects and given their detection rates. We find that:

• SDSS and Pan-STARRS1 together identified over half of the VHzQ sample;

• there remains a quasar ‘redshift desert’ at z ≈ 5.3−5.7, though efforts are being made to address this (e.g. Yang et al. 2018a);

• 96.9 per cent of the VHzQ sample is detected in one or more NIR (ZYJHK/K_s) band;

• the 15 objects that are not detected in the NIR are due to lack of coverage rather than lack of depth;

• 389 (79.7 per cent) VHzQs are detected by WISE, e.g. in the deeper unWISE W1 catalogue.

• All of the z ≥ 7 quasars are detected in both unWISE W1 and W2.

• 28 of the quasars had enough NIR measurements and sufficient NIR measurements and SNR to look for variability. Weak variability was detected in multiple bands of SDSS J0959+0227, and very marginally in the Y-band of MMT J0215–0529. Only one quasar, SDSS J0349+0034 had significant differences between their WFCAM and VISTA magnitudes in one band, indicating variability.

The science reach of z > 5 quasars will continue to be important well into the next decade (Becker et al. 2019; Wang et al. 2019b; Fan et al. 2019a) and will provide key insights into direct-collapse black holes, hydrogen reionization and the physics of accretion in the first ≲700 million years of the Universe.

4.1 Author Contributions

NPR initiated the project, compiled the list of z > 5.00 quasars, wrote most of the analysis code, developed the plotting scripts, and developed and wrote the initial and subsequent drafts of the manuscript.

NJGC supplied the critical NIR expertise and data base for which the bulk of the project relies. NJGC also contributed directly to the writing of the manuscript.

4.2 Availability of Data and computer analysis codes

All materials, databases, data tables and code are fully available at: https://github.com/d80b2t/VHzQ

SUPPORTING INFORMATION

Table 3. The first 24 (i.e. 5%) of 488 very high-z quasars in Right Ascension order with near and MIR photometry.

Please note: Oxford University Press is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

ACKNOWLEDGEMENTS

NPR acknowledges support from the STFC and the Ernest Rutherford Fellowship scheme. NJGC was supported by STFC grants to the Wide Field Astronomy Unit (ST/N005813/1 and ST/T000023/1).

We thank:

(i) Mike Read at the ROE WFAU for help with the WFCAM Science Archive (WSA) and the VISTA Science Archive (VSA).

(ii) Aaron Meisner and Eddie Schlafly for facilitating early access to the unWISE Catalogue.

(iii) Krisztina Perger for key discussions about missing objects after the pre-print appeared.

(iv) Tim Brooke at the IRSA Help Desk;

(v) Bernie Shiao at STScI for help with the Pan-STARRS1 DR1 CasJobs interface.

(vi) Michael Cushing for supplying the Late Type stellar spectra and Beth Biller for useful discussion.

(vii) Nathan Secrest for useful discussions on the MIR AGN catalogues.

This paper heavily used topcat (v4.4) (Taylor 2005, 2011). This research made use of Astropy, a community-developed core python package for Astronomy (Astropy Collaboration et al. 2013; The Astropy Collaboration et al. 2018).

The VISTA Data Flow System pipeline processing and science archive were used for the WFCAM and VISTA NIR data are described in Irwin et al. (2004), Hambly et al. (2008) and Cross et al. (2012).

The Pan-STARRS1 Surveys (PS1) and the PS1 public science archive have been made possible through contributions by the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, the Queen’s University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under Grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation Grant No. AST-1238877, the University of Maryland, Eotvos Lorand University (ELTE), the Los Alamos National Laboratory, and the Gordon and Betty Moore Foundation.

This project used data obtained with the Dark Energy Camera (DECam) and the NOAO Data Lab, The Data Lab is operated by the National Optical Astronomy Observatory, the national center for ground-based nighttime astronomy in the United States operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation.

This publication makes use of data products from the WISE, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, and NEOWISE, which is a project of the Jet Propulsion Laboratory/California Institute of Technology. WISE and NEOWISE are funded by the National Aeronautics and Space Administration.

CasJobs was originally developed by the Johns Hopkins University/ Sloan Digital Sky Survey (JHU/SDSS) team. With their permission, MAST used version 3.5.16 to construct CasJobs-based tools for GALEX, Kepler, the Hubble Source Catalogue, and PanSTARRS.

This research has made use of the SVO Filter Profile Service (http://svo2.cab.inta-csic.es/theory/fps/) supported from the Spanish MINECO through grant AyA2014-55216 The SVO Filter Profile Service¹⁰ describes the Spanish VO Filter Profile Service. The Filter Profile Service Access Protocol. Rodrigo, C., Solano, E. http://ivoa.net/documents/Notes/SVOFPSDAL/index.html

Footnotes

REFERENCES

APPENDIX A: NIR WFCAM SCIENCE ARCHIVE SQL QUERIES

Here, we give the recipe and SQL that returned the NIR photometry for the VHzQs from the WFCAM Science Archive.

The data are on the WFCAM Science Archive: wsa.roe.ac.uk. Access the User Login form wsa.roe.ac.uk/login.html with these credentials::

• Username: WSERV1000

• password: highzqso

• community: nonsurvey

Then going to the Free Form SQL Query page the Database release WSERV1000v20200113 can be accessed which contains all the data we use here.

We nota bene a few things. First, the quantity aperJky3 and aperJky3Err are found in the wserv1000MapRemeasAver and wserv1000MapRemeasurement, so care has to be taken to return unique column names (otherwise e.g. astropy.io.fits will crash). As such, we alias aver.aperJky3 to aperJky3Aver and likewise for the error quantity. Aliases will be necessary in some cases anyway, because some queries can be done sensibly on multiple instances of the same table. Other times, one may join tables on quantities such as catalogueID or apertureID, where you are meaning the same thing, but aliases are then necessary for SQLServer to correctly comprehend the query.

Second, the RA and DEC values returned by the WSA are in radians, if used directly. To return values in degrees, use a selection with an alias, e.g. RA as RADeg and DEC as DECDeg. Other relevant query examples can be found in the GitHub repository, under WSA_VSA/SAMPLE_SQL_QUERIES, or the SQLCookbook in the WSA or VSA.

A1 The following SQL will return the best magnitude in each band for each QSO

1 SELECT

2 qso. qsoName , qso .ra as raJ2000 , qso .dec as decJ2000 ,

3 aver . apertureID , aver . aperJky3 as aperJky3Aver ,

4 aver . aperJky3Err as aperJky3AverErr , aver . sumWeight ,

5 aver . ppErrBits as ppErrBitsAver , m. mjdObs ,

6 m. filterID , remeas . aperJky3 ,

7 remeas . aperJky3Err ,

8 w.weight , remeas . ppErrBits ,

9 m. project

10

11 FROM

12 finalQsoCatalogue as qso ,

13 MapApertureIDshighzQsoMap as ma ,

14 wserv1000MapRemeasAver as aver ,

15 wserv1000MapRemeasurement as remeas ,

16 MapProvenance as v,

17 wserv1000MapAverageWeights as w,

18 MapFrameStatus as mfs ,

19 Multiframe as m

20

21 WHERE

22 qso . qsoID = ma. objectID and

23 ma. apertureID = aver . apertureID and

24 aver . apertureID = remeas . apertureID and

25 aver . catalogueID = v. combicatID and

26 v. avSetupID = 0 and

27 v. catalogueID = remeas . catalogueID and

28 w. combicatID = v. combicatID and

29 w. catalogueID = v. catalogueID and

30 w. apertureID = aver . apertureID and

31 mfs . catalogueID = remeas . catalogueID and

32 m. multiframeID = mfs. multiframeID and

33 mfs . programmeID = 10999 and

34 mfs . mapID = 1

35 order by v. combicatID , m. mjdObs

APPENDIX B: NIR VISTA SCIENCE ARCHIVE SQL QUERIES

In a very similar manner to the WSA, we give here the details on how to access the VISTA Science Archive (VSA)

At the VSA Login, enter with these credentials::

• Username: VSERV1000

• password: highzqso

• community: proprietary

Then the Freeform SQL Query page is where the data base release to use is VSERV1000v20200113. The queries are the same as those used in the WSA, see Appendix  A, but the names of the tables starting with wserv1000 should be replaced with vserv1000.

APPENDIX C: UPDATED VHzQ SAMPLE

Between the journal submitted version of this paper (which was also the arXiv v1) we updated the sample of VHzQs from a total of 463 objects to 488 objects. Total 16 of these objects were from Yang et al. (2018a) and 9 objects were added after discussions with Krisztina Perger, with the relevant catalogue found at astro.elte.hu/∼perger/catalog.html. We do not include the object DES0141-54 (Belladitta et al. 2018), since we were unable to obtain arcsecond level coordinates to perform our matched-aperture photometry. We also update the positions of five objects, see Table C1.

One object in the Perger catalogue, VA-13492 we could not immediately cross-match. In this case the coordinates looked similar to SDSSJ1040–1155, and the stated redshift was identical, z = 5.44. Additionally the i-band magnitude was equivalent to SDSS J1040–1155 (i = 25.9 mag). SDSS J1040–1155 is first reported in Douglas et al. (2007) and a transcription error seems to have occurred in Flesch (2015).