WISE DETECTIONS OF KNOWN QSOs AT REDSHIFTS GREATER THAN SIX

At the highest redshifts, QSOs provide excellent tools for probing the process and end of reionization and provide signposts to some of the most active locations and phases in the process of galaxy evolution. A variety of surveys has been used to select active galaxies at high redshifts on the grounds of their optical and near-infrared (IR) colors. The optical Sloan Digital Sky Survey (SDSS; Abazajian et al. 2009) has mapped one-quarter of the sky, from which the high-redshift QSOs have been selected and studied by Fan et al. (2001, 2003, 2004, 2006) and Jiang et al. (2008). A deeper, narrower optical–IR QSO survey has also been conducted by the Canada–France–Hawaii Telescope (CFHT) Legacy Survey (e.g., Willott et al. 2007, 2010). Combining near-IR imaging data in the selection criteria, the UKIDSS survey currently holds the QSO redshift record at z = 7.086 (Mortlock et al. 2011). The southern 2dF and 6dF surveys (Croom et al. 2004; Heath Jones et al. 2009), based on digitized optical plate imaging and the IR Two Micron All-Sky Survey (2MASS), respectively, provide a substantial increase in the area surveyed for color-selected spectroscopically confirmed QSOs beyond the footprint of the SDSS. One further optically faint QSO was independently identified in the 8.5 deg² Spitzer Deep, Wide-Field Survey in Boötes by McGreer et al. (2006) and Stern et al. (2007). A current summary of references to z > 6 QSOs can be found in Wang et al. (2011b). Deep observations using Spitzer have confirmed that at least some of the highest-redshift QSOs are detectable in the mid-IR (Charmandaris et al. 2004; Hines et al. 2006; Jiang et al. 2006; Stern et al. 2007), with properties not dramatically different than lower-redshift examples (Elvis et al. 1994; Staguhn et al. 2005). Some high-redshift QSOs are certainly also very rich in gas and dust, and provide most of the significant detections of ultraluminous galaxies in molecular lines at the highest redshifts (Wang et al. 2011a; Maiolino et al. 2012). There appears to be significant variation in the amounts of both the hottest and the total dust present, however (see Jiang et al. 2010). Some relevant results of analyzing spectral energy distributions (SEDs) of high-redshift active galaxies based on wideband UV–radio observations, including data from Spitzer and Herschel, are presented in Leipski & Meisenheimer (2012); in particular, the long-wavelength far-IR and submillimeter properties of high-redshift active galaxies are discussed by Leipski et al. (2013).

The recent spectroscopic confirmation by Mortlock et al. (2011) of the first bright near-IR-selected QSO at z > 7 spurred a search for the properties of known QSOs at high redshifts in the WISE All-Sky survey (Wright et al. 2010). WISE has mapped the whole sky in four IR bands, centered at 3.4, 4.6, 12, and 22 μm (bands W1–W4, respectively), with the most relevant bands W1 and W2 reaching a typical 5σ sensitivity of 6.8 and 9.8 μJy, respectively, and an angular resolution of about 6 arcseconds. As a result of the polar orbit of WISE, depths vary across the sky—there is deeper coverage at the poles than at the equator.

We compiled a list of all 31 published spectroscopically confirmed QSOs at z > 6 and searched the WISE All-Sky data release products catalog (2012 March 14; Cutri et al. 2012)¹¹ to find detections and limits of these objects in the WISE bands from 3.4 to 22 μm. The ≃ 6 arcseconds spatial resolution of WISE is approximately three times coarser than the 1.7 arcseconds available from Spitzer-IRAC at 3.6 μm, and the integration time per point on the sky from WISE is typically only two minutes. However, the images have excellent astrometry and a search of the WISE database provides sensitive mid-IR (>5 μm) all-sky imaging, reaching many times deeper than existing surveys from IRAS and Akari. Seven of our targets have archival detections by Spitzer (Jiang et al. 2006; Stern et al. 2007), but most were published after the cryogenic phase of Spitzer was completed, and so WISE provides the only source of mid-IR information longward of 5 μm on most of these targets until the launch of James Webb Space Telescope (JWST). Most of the detections are made in WISE bands W1 and W2, which approximately match the IRAC [3.6] and [4.5] bands that are still operating in the Spitzer warm mission. Table 1 presents 17 WISE detections, 14 in the main all-sky release source catalog with >5σ detections and 3 from the "Reject Table"¹² that contains >3.5σ detections. This confirms that Spitzer can easily continue to detect the highest-redshift QSOs while the warm mission continues.

High-redshift QSOs are typically selected on the grounds of their smooth, broad, power-law spectra redward of a strong redshifted Lyman-break feature that at these redshifts lies between the i and z bands for z ∼ 6 and between the z and J bands for the z = 7.02 UKIDSS QSO. The WISE bands alone provide little additional power to select these objects, owing to the relatively featureless continuum emission of active galactic nuclei (AGN) yielding similar colors across the WISE bands at all redshifts (see Figure 1). However, when WISE detects high-redshift AGN, the measured fluxes provide a valuable measure of the rest-frame optical spectral energy distribution in the W1 and W2 bands and signs of hot dust emission from the immediate environment of the AGN can be studied using the W3- and W4-band data.

Zoom In Zoom Out Reset image size

In Section 2, we discuss the WISE properties of the z > 6 QSOs, and their multiwavelength properties in Section 3. The quoted magnitudes are in their native system (Vega for WISE, AB for SDSS) unless otherwise stated (note Figures 5 and 6).

The list of 17 z > 6 QSOs that have cataloged WISE detections are presented in Table 1, along with their flux densities or limits from the WISE catalogs. Of the 14 all-sky release detections in W1, 13 are also detected in the W2 band; a subset of three are also listed as detections in W3; none are detected in W4. All of the reported detections—which are distinguished in the table by not having a null (...) entry for a noise estimate—have a WISE pipeline signal-to-noise (S/N) ratio value (wXsnr) greater than 3.5 for band X = 1, 2, or 3. The WISE colors of these high-redshift QSOs are compared with a match of the WISE and the SDSS QSO catalogs (Schneider et al. 2007) in Figure 1. That match yielded WISE colors for 19,876 QSOs at z < 5.6.

The WISE-detected high-redshift QSOs are all found at similar positions in the WISE W1–W2:W2–W3 color space, close to (3.0, 1.0), deep within the densest region of WISE colors for lower-redshift AGN (Figure 1). This highlights the difficulty of using WISE photometry alone to identify candidate high-redshift QSOs. Where there are only WISE limits, the regions where QSOs may lie covers almost the whole of this color space. Combining WISE and optical/near-IR data to highlight candidates is more promising, as we discuss in Section 3.

The WISE depth of coverage varies across the sky, with depth increasing away from the ecliptic plane toward the poles. The detectability of a target QSO appears to be related to this factor at least in part. The number of individual imaging frames taken by WISE that were included in the creation of the WISE Atlas Images for the 14 detections in the All-Sky Release source catalog ranged from 11–48, with a median of 21 and an interquartile range of 13–28. These numbers are unchanged when the three Reject-Table QSOs are included. For the 14 WISE non-detections, the number of individual WISE 8 s exposures that contributed to the catalog entry was 12–24, with a median of 15 and an interquartile range of 13–22. We discuss the results of stacking a total of 188 of these 8 s single-exposure WISE image frames for the 10 non-detected sources below and conclude that, based on their different WISE fluxes, this coverage effect alone is unlikely to account for the difference between the detection and non-detection of high-redshift QSOs by WISE.

The WISE Atlas Images corresponding to the high-redshift QSOs were inspected to ensure that the reported source was not affected by image artifacts or other problems. ULAS J1120+0641 is within 15 arcseconds of a bright star; however, the photometry is uncontaminated. The images for the undetected z > 6 QSOs were also inspected, and while in some cases, as expected, net positive signals are found at the QSO position, they are all consistent with a non-detection and thus are correctly reported as non-detections in the pipeline-processed WISE catalog. Further data was taken in the W1 and W2 bands after the cryogen was exhausted, the "WISE Post-Cryo Survey Phase" of the mission. These data will be combined with the cryogenic mission data in the near future, resulting in an "All-WISE" catalog with an anticipated noise level about 1.4 times deeper over about 70% of the sky, increasing the utility of WISE for detecting QSOs samples further.

Five QSOs—SDSS J1030+0524, SDSS J1148+5251, QSO J1250+3130, QSO B1425+3326, and SDSS J1602+4228—have bright W1 and W2 detections. Their WISE images, along with the images of the highest-redshift target ULAS J1120+0641, are shown in Figure 2.¹³ In the case of SDSS 1030+0524, SDSS J1148+5251, and QSO B1425+3326, there are Spitzer measurements (Jiang et al. 2006; Stern et al. 2007) that are consistent with the WISE W1 and W2 flux densities reported in Table 1. Three have a relatively large number of detections, whereas QSO J1250+3130 is a very typical source.

Zoom In Zoom Out Reset image size

In order to investigate the non-detections further, both for the W1 and W2 bands in the non-detected targets and for the W3 and W4 bands in the detected targets, we stacked the WISE images from the QSO positions. In order to overcome the variations in the coverage of different targets, the whole ensemble of 8 s duration single-exposure WISE images that overlapped the target positions were extracted from the WISE Image Atlas and stacked with hot pixels and bright sources masked in order to probe the net flux density of the sample. This was done for all the targets without nearby bright objects, corresponding to 300 W1 frames, 300 W2 frames, 297 W3 frames, and 297 W4 frames for 16 (of 17 in total) high-redshift QSOs with detections in the bluer WISE bands. Stacking the frames for the 10 sources uncontaminated by nearby brighter objects (of 14 in total) without WISE detections in W1 or W2 provided a total of 188 frames in each of these bands. These stacked images are shown in Figure 3, with the same grayscale display range as shown in Figure 2. Note that the pixel scale in the W4 images is twice that in the other three bands. There is reasonably clear detection of the WISE-cataloged objects in the stack in both the W1 and W2 bands at S/N ratios of 13.5 and 10.9, respectively. By comparison, the stack of the non-detected QSOs shows no sign of emission in the W1 or W2 bands (Figure 3), with a noise level in the stacked image that is 26% greater, reflecting the lesser number of uncontaminated fields included in the stack of non-detected targets (188 versus 300). There are no indications of any net detections in the longer WISE bands, W3 and W4, for any subsamples, which is reasonable given the sensitivity and the modest number of stacked fields.

Zoom In Zoom Out Reset image size

Taken at face-value, this indicates that the fraction of Type-1 QSOs at z > 6 that are not detected in the WISE catalog are systematically and substantially fainter in WISE bands W1 and W2 than the WISE-detected QSOs; they are not just a little fainter with the non-detections just evading detection below a formal flux limit. We now estimate the relative brightness of the WISE-detected and WISE-non-detected QSOs. We combine the results in bands W1 and W2 based on the results of the stacks described above. The average S/N ratio of the WISE-detected QSOs is 12.2 (the mean of 13.5 and 10.9) and the relative noise level in the stack of the non-WISE-detected QSOs is 1.26. Taking a 3σ limit for the WISE non-detected sources, this gives a flux ratio of 12.2/(3 × 1.26) ≃ 3.2. We interpret this as meaning that the WISE-non-detected QSOs in the sample considered here are likely to be, on average, a factor of about three times fainter than the WISE-detected QSOs.

The number of galaxies in this stack is very modest, however, and so the statistical power is limited. Nevertheless, there does appear to be a real difference in the mid-IR flux densities of the WISE-detected and WISE-non-detected z > 6 QSO targets by an amount comparable to the offsets in their rest-UV optical magnitudes discussed further in Section 3.

We now discuss the WISE properties of the z > 6 QSOs in the context of their optical and near-IR emission. The components of the emission that contributes to the brighter WISE-detected examples are investigated using SED models in the WISE bands in Figure 4. They are not significantly different from those previously reported at low and modest redshifts for Spitzer-detected QSOs (Richards et al. 2006; Stern et al. 2012). Deep Spitzer observations of high-redshift QSOs (Jiang et al. 2010) have revealed some QSOs at z ∼ 6 with little or no hot dust emission in the 24 μm band. The larger number of much shallower observations by WISE in the W3 (12 μm) band, which is sensitive to the hottest dust, could eventually be stacked to help to reveal whether these objects are unusual or typical. At face value, our z > 6 sample indicates no obvious difference in the shapes of the SEDs of the WISE-detected and WISE-non-detected z > 6 QSOs considered here to the depth of the survey.

Zoom In Zoom Out Reset image size

There seems to be little possibility that the WISE colors could be affected dramatically by redshifted line emission. In the range 6 < z < 7, Lyα lies in the i and z bands, while Hα is in WISE W2 and Hβ is in WISE W1. The average restframe equivalent width is 194 Å  and 46 Å  for Hα and Hβ from the SDSS QSO template of vanden Berk et al. (2001). Redshifting these lines from z = 6.5 would correspond to equivalent widths of 0.15 μm for Hα in W2 and 0.035 μm for Hβ in W1, respectively. The fractional bandwidths are approximately 25%, and so about 15% of the W2 flux and 5% of the W1 flux might be expected to be contributed by these lines. Although there is evidence that the equivalent width of Hα increases in galaxies at higher redshifts (Shim et al. 2011), this line contamination effect is unlikely to be significant here in light of the size of the measurement uncertainties.

The WISE colors of the z > 6 detections are compared with the WISE colors of larger samples of lower-redshift QSOs in Figure 1. The WISE–SDSS cross-matched sample has much more complete coverage of low-redshift objects than other surveys. The SDSS sample contains relatively few lower-redshift QSOs around (4.2, 0.4), in the region of WISE color space where the z > 6 QSOs are found (Figure 1); it includes a larger, more complete sample of QSOs with matches in the WISE catalog at 3 < z < 5 and extends to a maximum redshift of z = 5.41. At modest redshifts, z < 2–3, the WISE colors provide a selection comparable to that described by Stern et al. (2005) for more moderate redshifts due to the relative differences between the colors of evolved stellar light and the AGN power law that dominates at longer wavelengths, as revealed by the W1−W2 color (see Stern et al. 2012; the discussion below and Figure 4). However, despite the confirmed ability of WISE to detect QSOs out to the end of the epoch of reionization, we emphasize that the colors of the WISE-detected objects do not stand out from the much larger numbers of AGN at more moderate redshifts (Figure 1), as their colors are determined by a broken power-law SED that does not change very strongly with increasing redshift. Thus the discovery of purely WISE-selected samples of high-redshift AGN are unlikely; there is no unambiguous signature of a very high redshift QSO (z > 6) from the WISE data alone at the WISE depth. The W1−W2 colors of the z > 6 QSOs are typically bluer than those at low redshifts, as expected, but they are difficult to distinguish from z ≃ 0.5 star-forming galaxies as many are bluer than the W1−W2 >0.8 color cut imposed in WISE color-based AGN selection methods (see Stern et al. 2012; Mateos et al. 2012; Assef et al. 2013).

Furthermore, most of the W3 flux values are upper limits and so the WISE points without error bars can move to the left in the W2−W3 color space of Figure 1, corresponding to bluer mid-IR colors. However, the detected z > 6 QSOs remain within the cloud of lower-redshift QSOs and luminous IR galaxies (Wright et al. 2010; Jarrett et al. 2011) in this color space, preventing any easy sifting of QSO candidates from the WISE data.

WISE data add significantly to the available information about high-redshift candidates, as shown in Figures 4–6. The SEDs of four of the brighter WISE detections, whose images are shown in Figure 2, are compared with templates from Assef et al. (2010) in Figure 4. These fits allow an estimate of the relative SED contributions of an evolved stellar component and an AGN component. The AGN is described approximately by two power laws, with a break near 1 μm (e.g., vanden Berk et al. 2001). These templates appear to provide a reasonable description of the SEDs of the z > 6 QSOs. In no case is there compelling evidence that a large fraction of light from a young star-forming population (Sbc/Im) is required in the models to account for the data. Less precise SED information for more of the WISE-detected and non-detected QSOs is shown in Figure 6.

Seven of the 17 WISE-detected z > 6 QSOs listed in Table 1 have Spitzer detections (see the footnotes to Table 1). They typically have a median of 3 bands of iz JHK photometry available, while the 14 WISE non-detected z > 6 QSOs have a median of 2 detections in these bands.

The relative rest-UV colors of the WISE-detected and WISE-non-detected z > 6 QSOs are compared in Figure 5. The relative magnitudes and colors, indicating that the WISE non-detections are both fainter by about two magnitudes in i and z, have a modestly greater range in their z-band magnitudes, and are marginally bluer in i − z by about half a magnitude. As discussed above, i − z straddles the wavelength of Lyα at these redshifts and so this rest-UV color could be affected strongly by the column density of external and internal hydrogen absorption and the effects of extinction. Note that the highest-redshift example, ULAS J1120+0641, does not appear in this plot as Lyα has redshifted through both of the relevant bands at z > 7. These differences are consistent with the WISE detectability amongst the 31 z > 6 QSOs being related to the fainter z-band emission from the targets not detected by WISE. If the rest-UV light comes from accretion (Figure 4) and is, on average, two magnitudes fainter in the WISE-non-detected sample than the WISE-detected one, then it could just reflect significantly different accretion powers between the subsamples without any major difference in the optical–mid-IR colors. Note that there is inadequate deep near-IR data (see also Figure 6) to construct comparable histograms for the observed near-IR colors/magnitudes for these samples. Some information is available from deeper observations (e.g., Staguhn et al. 2005) and all the available information is included in the SEDs shown in Figure 6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The large number of AGN candidates that are sure to be detected using PanSTARRS (Morganson et al. 2012), SkyMapper, VISTA, VST, DES, and LSST can be matched to the WISE catalog in order to identify particularly dusty examples, to get a better idea of their true luminosities, and, at the highest redshifts, to obtain more useful limits on their stellar masses. The significant detection rate of existing z > 6 QSO samples in the WISE All-Sky Release catalog, mostly from within the SDSS footprint, and in the deeper, smaller area CFHT surveys offers a potentially large sample of future candidates. The WISE detection of a substantial fraction of known z > 6 QSOs, and one at z > 7, makes us optimistic that candidates at even higher redshifts could be found and isolated by taking advantage of a combination of deep wide-field optical–near-IR surveys and the WISE catalog. When WISE detects a QSO and a redshift is known, the WISE data also immediately provides limits to its hot dust emission from W3, to its rest-optical luminosity from W1 and W2, and even perhaps an evolved stellar population, all of which would otherwise be difficult to obtain until the launch of JWST.

We have searched for 31 known z > 6 QSOs in the WISE All-Sky survey and detected 17/31 (55%) in band W1, and 16, 3, and zero in bands W2, W3, and W4, respectively. This significant detection rate for QSOs selected from both SDSS and deeper imaging should make WISE a useful resource for exploiting ongoing wide-optical/near-IR imaging surveys. The SED properties of the brighter WISE-detected objects are not unlike those of lower-redshift samples. From stacking the non-detected sample of 14 QSOs, their 3.4 and 4.6 μm emission (in bands W1 and W2, respectively) appears to be significantly less powerful than that of the WISE-detected QSOs, indicating that there is likely to be a significant difference between the mid-IR luminosities of the two classes, with WISE-non-detection due to intrinsic faintness and not simply due to a matter of a chance appearance above the noise level in observations with low S/N ratios.

We thank the anonymous referee for a careful reading and helpful suggestions. This publication makes use of data products from the Wide-field Infrared Survey Explorer (WISE). WISE is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration. This research has made use of both the NASA/IPAC Infrared Science Archive (IRSA) and the NASA/IPAC Extragalactic Database (NED), which are operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. R.J.A. was supported by an appointment to the NASA Postdoctoral Program at the Jet Propulsion Laboratory, administered by Oak Ridge Associated Universities through a contract with NASA.

Note added in proof. The enhanced AllWISE data release is scheduled for late 2013. The AllWISE catalogs are being compiled as this paper goes to press. Of the sources listed in Table 1, 16 are in the new source tables. The exception is SDSS J2054−0005, which was listed in the All-sky Reject Table. Searching the 31 high-redshift targets in the new tables yields one additional target that satisfies the conditions for inclusion in the All-Sky Release Catalog and three additional targets that satisfy the conditions for being entered in the All-Sky Reject Table, while two sources are promoted from the Reject Table to the catalog and one source is demoted from the catalog to the Reject Table. Hence, the 14 targets listed in the All-Sky Release Catalog and the 3 in the Reject Table in Table 1 for the WISE All-Sky data release are likely to be changed to 16 and 4, respectively, in the AllWISE Catalog, giving an increased cataloged fraction of 65% for known QSOs at redshifts greater than six.

Facility: WISE - Wide-field Infrared Survey Explorer