The Discovery of a Highly Accreting, Radio-loud Quasar at z = 6.82

P172+18 has been identified as a z > 6.5 quasar candidate by at least two independent methods. We first selected P172+18 as a z-dropout radio-loud candidate in Bañados et al. (2015) (see their Table 1). That selection required red (z_P1 − y_P1 > 1.4) sources in the stacked object Pan-STARRS1 catalog (Chambers et al. 2016) and a counterpart in the radio survey Faint Images of the Radio Sky at Twenty cm (FIRST, Becker et al. 1995) to avoid most of the L- and T-dwarfs, which are the main contaminants for z > 6.5 quasar searches (see Bañados et al. 2015 for details). The Pan-STARRS1 and FIRST measurements for P172+18 are listed in Table 1. This object also stands out as a promising high-redshift quasar candidate in a new method to select z ≳ 6.5 quasars exploiting the overlap of Pan-STARRS1 and the DESI Legacy Imaging Surveys (DECaLS; Dey et al. 2019), which will be presented in a forthcoming paper along with additional z ≳ 6.5 quasar discoveries (E. Bañados et al. in preparation). P172+18 was selected using the DECaLS DR7 catalog, but in Table 1 we report the photometry from the latest (DR8) data release.

Table 1. Photometry of the Radio-loud Quasar P172+18 and its Radio Companion

	Quasar	Radio Companion
R.A. (J2000)	11h29m2537	11h29m2408
Decl. (J2000)	$+18^\circ {46}^{{\prime} }24\buildrel{\prime\prime}\over{.} 29$	$+18^\circ {46}^{{\prime} }38\buildrel{\prime\prime}\over{.} 58$
Public optical and infrared surveys
Pan-STARRS1 iP1	>23.6	>23.6
Pan-STARRS1 zP1	>23.2	>23.2
Pan-STARRS1 yP1	20.76 ± 0.09	>22.3
DECaLS DR8 gDE	>25.4	>25.4
DECaLS DR8 rDE	>24.8	>24.8
DECaLS DR8 zDE	21.64 ± 0.05	>23.8
DECaLS DR8 W1	20.71 ± 0.13	>21.8
DECaLS DR8 W2	20.73 ± 0.31	>20.9
Follow-up near-infrared imaging
JNOT	20.90 ± 0.11	>22.2
HNOT	21.36 ± 0.24	>21.8
KsNOT	21.07 ± 0.18	>21.7
Public radio surveys
TGSS 147.5 MHz	<8.5 mJy	<8.5 mJy
FIRST 1.4 GHz	1020 ± 144 μJy a	<406 μJy
Radio follow-up
VLA-L 1.52 GHz	510 ± 15 μJy	732 ± 15 μJy
VLA-S 2.87 GHz	222 ± 9 μJy	432 ± 20 μJy b
${\alpha }_{S}^{L}$	−1.31 ± 0.08	−0.83 ± 0.08
Quasar rest-frame luminosities c
m1450	21.08 ± 0.10
M1450	−25.81 ± 0.10
L2500	(1.4 ± 0.1) × 1046 erg s−1
L3000	(1.3 ± 0.1) × 1046 erg s−1
L4400	(1.1 ± 0.1) × 1046 erg s−1
L5 GHz	(5.4 ± 0.2) × 1042 erg s−1

Notes.

^a This is the reported peak flux density in the FIRST catalog (version 2014dec17). We note that in the FIRST image we measure 852 ± 135 μJy. ^b The source is marginally resolved in the VLA-S image and we report the integrated flux. ^c The quasar UV and optical luminosities are derived from the best-fit power law of the near-infrared spectrum (see Table 4) and the uncertainties are dominated by the J_NOT photometry used for absolute flux calibration of the spectrum. The 5 GHz radio luminosity is extrapolated using the measured radio index.

Download table as:  ASCIITypeset image

The optical photometry of P172+18 in the DECaLS DR7 and DR8 catalogs is consistent. However, the mid-infrared Wide-field Infrared Survey Explorer (WISE) magnitudes are inconsistent ²¹ at the 2σ level even though DR7 and DR8 use the same input set of WISE images spanning from 2010 to 2017 (A. Meisner, private communication; Meisner et al. 2019). DECaLS provides matched WISE photometry by using the g_DE, r_DE, z_DE information to infer the WISE magnitudes from deep image coadds using all available WISE data (Lang 2014; Meisner et al. 2017, 2019). The WISE DECaLS DR7 magnitudes are W1 = 21.25 ± 0.21 and W2 = 21.30 ± 0.51 in contrast to the DR8 magnitudes of W1 = 20.71 ± 0.13 and W2 = 20.73 ± 0.31. The main difference between DR7 and DR8 is the change of sky modeling as presented in Schlafly et al. (2019), which can affect the fluxes of sources at the faint limit of the unWISE coadds. Therefore, the reported WISE magnitudes need to be taken with caution.

We confirmed P172+18 as a z ∼ 6.8 quasar on 2019 January 12 with a 450 s spectrum using the Folded-port Infrared Echellete (FIRE; Simcoe et al. 2008, 2013) spectrograph in prism mode at the Magellan Baade telescope at Las Campanas Observatory. The spectrum had poor signal-to-noise ratio (S/N) but was sufficient to unequivocally identify P172+18 as the most distant radio-loud quasar known to date, which triggered a number of follow-up programs described below.

We obtained JHK photometry using the NOTCam instrument at the Nordic Optical Telescope (Djupvik & Andersen 2010). The total exposure times were 19 minutes each for J_NOT and H_NOT and 31 minutes for Ks_NOT. Data reduction consisted of standard procedures: bias subtraction, flat-fielding, sky subtraction, alignment, and stacking. Table 2 presents a log of the observations.

We calculate the zero-points of the NOT observations, calibrating against stars in the Two Micron All Sky Survey (2MASS) using the following conversions:

$$\begin{eqnarray*}\begin{array}{rcl}{J}_{\mathrm{NOT}} & = & {J}_{2\mathrm{MASS}}-0.074\times ({J}_{2\mathrm{MASS}}-{H}_{2\mathrm{MASS}})+0.003\\ {H}_{\mathrm{NOT}} & = & {H}_{2\mathrm{MASS}}+0.045\times ({J}_{2\mathrm{MASS}}-{H}_{2\mathrm{MASS}})+0.006\\ {{Ks}}_{\mathrm{NOT}} & = & {K}_{2\mathrm{MASS}}+0.580\times ({H}_{2\mathrm{MASS}}-{K}_{2\mathrm{MASS}})+0.225.\end{array}\end{eqnarray*}$$

These conversions were calculated via linear fits of the stellar loci as described in Section 2.6 of Bañados et al. (2014). The near-infrared photometry is listed in Table 1.

We obtained three follow-up spectra of P172+18. On 2019 February 18 we observed P172+18 for 3.5 hours with Keck/NIRES (Wilson et al. 2004). Between 2019 March 8 and April 8 we used the Very Large Telescope (VLT)/X-Shooter spectrograph (Vernet et al. 2011) to observe the target for a total time of 3.5 hr. We also observed P172+18 with the Large Binocular Telescope (LBT)/Multi-Object Double Spectrographs (MODS) (Pogge et al. 2010) on 2019 June 13. The MODS observations were carried out in binocular mode for 20 minutes on-source. We summarize the spectroscopic follow-up observations in Table 3.

Table 3. Summary of the Optical and Near-infrared Follow-up Spectroscopic Observations

Date	Telescope/Instrument	Exposure Time	Wavelength Range	Slit Width
2019 Feb 18	Keck/NIRES	3.5 hr	9400–24000 Å	055
2019 Mar 8–Apr 8	VLT/X-Shooter	3.5 hr	3000–24800 Å	10/09/06
2019 Jun 13	LBT/MODS	0.3 hr	5000–10000 Å	122

Download table as:  ASCIITypeset image

The Keck/NIRES and VLT/X-Shooter data were reduced with the Python Spectroscopic Data Reduction Pipeline (PypeIt; Prochaska et al. 2019, 2020). In practice, sky subtraction on the 2D images was obtained through a B-spline fitting procedure and differences between AB dithered exposures. The 1D spectrum was extracted with the optimal spectrum extraction technique (Horne 1986). Each 1D single exposure was flux-calibrated using standard stars observed with X-Shooter. Then, the 1D spectra were stacked and a telluric model was fitted, obtained from telluric model grids from the Line-By-Line Radiative Transfer Model (LBLRTM4; Clough et al. 2005; Gullikson et al. 2014). The X-Shooter and NIRES spectra were then absolute-flux-calibrated with respect to the J_NOT magnitude (see Table 1). The LBT/MODS binocular spectra were reduced with IRAF using standard procedures, including bias subtraction, flat-fielding, and telluric and wavelength calibration. They were each scaled to the y_P1 magnitude.

We performed all measurements presented in the following sections in the individual spectra, which resulted in consistent results. To maximize the information provided by all spectra we re-binned them to a common wavelength grid with a pixel size of 50 km s⁻¹, and averaged them weighting by their inverse variance. The final spectrum that we use for our main analysis is shown in Figure 1 and a zoom-in on the main emission lines is presented in Figure 2.

Follow-up radio-frequency observations were carried out with the Karl G. Jansky Very Large Array (VLA) of the NRAO ²² on 2019 March 5 and 2019 March 11, in S and L bands respectively. Each observing session was 1 hr in total (∼21 min on-source). The VLA was in B-configuration with a maximum baseline length of 11.1 km. The observations spanned the frequency ranges 1–2 GHz (L band; center frequency 1.5 GHz) and 2–4 GHz (S band; center frequency 3 GHz). The WIDAR correlator was configured to deliver 16 adjacent subbands per receiver band, each 64 MHz at L band and 128 MHz at S band. Each subband had 64 spectral channels, resulting in 1 MHz channels in the L-band data and 2 MHz channels in the S-band data.

The source 3C 286 (J1331+3030) was used to set the absolute flux density scale and to calibrate the bandpass response, and the compact source J1120+1420 was observed as the complex gain calibrator. Data editing, radio-frequency interference (RFI) excision, calibration, imaging, and analysis were performed using the Common Astronomy Software Applications (CASA) package of the NRAO. The data were calibrated using the CASA pipeline version 5.4.1–23, and the continuum images were made using the wide-field w-projection gridder and Briggs weighting with robust=0.4 as implemented in the CASA task tclean. Due to the excision of data affected by RFI, the resulting L- and S-band images have the reference frequencies of 1.52 and 2.87 GHz, respectively. The resulting beam sizes for the 1.52 and 2.87 GHz images are 355 × 324 and 227 × 185, respectively. A summary of the radio observations is listed in Table 2 and the results are discussed in Sections 4.1 and 4.2. The follow-up radio images as well as archival data from the FIRST survey are shown in Figure 3.

Specific properties such as equivalent width (EW) and peak velocity shift of key broad emission lines (e.g., Lyα, N v, C iv, and Mg ii) have been shown to trace properties of the innermost regions of quasars and of their accretion mechanisms (e.g., Leighly & Moore 2004; Richards et al. 2002, 2011).

We measure the redshifts of the emission lines as

$$\begin{eqnarray}&&{z}_{{\rm{line}}}=\displaystyle \frac{{\lambda }_{{\rm{line}},{\rm{obs}}}}{{\lambda }_{{\rm{line}},{\rm{rf}}}}-1\end{eqnarray} \tag{ 3 }$$

where λ_line,obs is the observed line wavelength, i.e., the peak of the fitted Gaussian function, and λ_line,rf is the rest-frame line wavelength (see Table 4). In case of a line fitted with two Gaussian functions (e.g., C iv and Lyα), we considered the peak wavelength corresponding to the maximum flux value of the full model (see Schindler et al. 2020 for further details).

P172+18 presents strong and narrow Lyα and N v emission lines (see Figures 1 and 2). We derive the total equivalent equivalent width of Lyα + N v, EW(Lyα + N v) ∼ 56 Å (see Table 4). This is consistent with the mean of the EW(Lyα + N v) distributions for 3 < z < 5 and z > 5.6 quasars as found by Diamond-Stanic et al. (2009) and Bañados et al. (2016), respectively. Notably for a z ∼ 7 quasar, the narrow component of the Lyα emission of P172+18 can be fitted well by a single Gaussian and there is no evidence for an IGM Lyα damping wing (see Wang et al. 2020), implying that the surrounding IGM is >90% ionized (see also Section 3.3).

Zoom In Zoom Out Reset image size

Now we focus on the relation between the C iv EW and the blueshift with respect to the Mg ii line. As a reminder, we model the C iv line with two Gaussians (see Table 4 and Figure 2). In the following, we consider all the components of the model, i.e., the total line emission. ²³ We measure C iv EW = ${21.3}_{-2.0}^{+2.4}$ Å and Δv_{Mg II−C IV} = 195 ± 225 km s⁻¹. In Figure 4, we place the measurements of P172+18 in the context of quasar populations at z ∼ 2 and z ≳ 6. For the z ∼ 2 subsample we select quasars from the Sloan Digital Sky Survey (SDSS) Data Release 7 quasar catalog (DR7; Shen et al. 2011) using the criteria of Richards et al. (2011):

• 1.  
  1.54 < z < 2.2, to ensure that both C iv and Mg ii emission lines are encompassed by the SDSS spectral wavelength range.
• 2.  
  FWHM_{C IV} and FWHM_{Mg II} > 1000 km s⁻¹, to select only quasars with broad emission lines.
• 3.  
  FWHM_{C IV} > 2σ_{FWHM,C IV} and EW_{C IV} > 2σ_{EW,C IV} and EW_{C IV} > 5 Å, for a reliable fit of the C iv line.
• 4.  
  FWHM_{Mg II} > 2σ_{FWHM,Mg II} and EW_{Mg II} > 2σ_{EW,Mg II}, for a reliable fit of the Mg ii line.
• 5.  
  we exclude broad absorption line quasars (BAL_FLAG = 0).

This yields 22,703 objects, out of which 1284 are classified as radio-loud with R₂₅₀₀ > 10.

Zoom In Zoom Out Reset image size

As shown in Richards et al. (2011), radio-loud quasars occupy a specific region of the C iv EW–blueshift parameter space: small blueshifts (≲1000 km s ⁻¹) but a wide range of EW values. However, note that for each radio-loud quasar several radio-quiet ones with similar rest-frame UV properties can be found, but not necessarily the other way around (Figure 4). Recently, the C iv emission line of z ≳ 6 quasars has been studied by various researchers (e.g., Mazzucchelli et al. 2017; Meyer et al. 2019). Large blueshifts for these objects are ubiquitous, with median values of Δv_{Mg II−C IV} ∼ 1800 km s⁻¹ (Schindler et al. 2020) and with extreme values extending to Δv_{Mg II−C IV} ≳ 5000 km s⁻¹ (e.g., Onoue et al. 2020). In Figure 4 we show the Δv_{Mg II−C IV} measurements for z > 6 quasars from Mazzucchelli et al. (2017), Shen et al. (2019), Onoue et al. (2020), and Schindler et al. (2020).

To exclude objects with extremely faint emission lines and/or with spectra with low S/N close to the C iv line, we consider only high-z quasars for which EW_{C IV} > 2σ_{C IV} and EW_{C IV} > 5 Å. Out of the three radio-loud quasars at z > 6 that have near-infrared spectra covering Mg ii and C iv, only J1429+5447 does not satisfy our criteria owing to its extremely weak emission lines (EW_{C IV} < 5 Å; Shen et al. 2019). The two radio-loud quasars at z > 6 in Figure 4, J1427+3312 and P172+18, show C iv emission line properties consistent with what is observed in the radio-loud sample at z ∼ 2. A larger sample of radio-loud quasars at high redshift with near-infrared spectra is needed to further investigate whether this trend changes with redshift, and whether the different EW and blueshift properties of radio-loud quasars can inform us about physical properties of their broad-line regions and/or their accretion mode.

We compute the quasar bolometric luminosity (L_bol) using the bolometric correction presented by Richards et al. (2006):

$$\begin{eqnarray}&&{L}_{\mathrm{bol}}=5.15\,\lambda \,{L}_{\lambda }(3000\,\mathring{\rm A} )\,\mathrm{erg}\,{{\rm{s}}}^{-1}\end{eqnarray} \tag{ 4 }$$

where L_λ (3000 Å) is the monochromatic luminosity at 3000 Å derived from the power-law model. We estimate the black hole mass using the Mg ii line as a proxy through the scaling relation presented by Vestergaard & Osmer (2009):

$$\begin{eqnarray}&&{M}_{{\rm{BH}}}={10}^{6.86}{\left[\displaystyle \frac{{\rm{FWHM}}({\rm{Mg}}{\rm\small{II}})}{{10}^{3}{\rm{km}}{{\rm{s}}}^{-1}}\right]}^{2}{\left[\displaystyle \frac{\lambda {L}_{\lambda }(3000\mathring{{\rm{A}}})}{{10}^{44}{\rm{erg}}{{\rm{s}}}^{-1}}\right]}^{0.5}\,{M}_{\odot }.\end{eqnarray} \tag{ 5 }$$

This scaling relation has an intrinsic scatter of 0.55 dex, which is the dominant uncertainty of the black hole mass estimate. Once we have a black hole mass estimate, we can directly derive the Eddington luminosity as

$$\begin{eqnarray}&&{L}_{\mathrm{Edd}}=1.3\times {10}^{38}\left(\displaystyle \frac{{M}_{\mathrm{BH}}}{{M}_{\odot }}\right)\,\mathrm{erg}\,{{\rm{s}}}^{-1}.\end{eqnarray} \tag{ 6 }$$

We obtain a black hole mass of ${M}_{{\rm{BH}}}={2.9}_{-0.6}^{+0.7}\times {10}^{8}\,{M}_{\odot }$ and an Eddington ratio of L_bol/${L}_{\mathrm{Edd}}={2.2}_{-0.4}^{+0.6}$ for P172+18 (see also Table 5). We note that the Eddington ratio depends on the bolometric luminosity correction used. For example, using the correction recommended by Runnoe et al. (2012),

$$\begin{eqnarray}&&{\rm{log}}{L}_{{\rm{bol}}}=1.852+0.975\times {\rm{log}}(\lambda \,{L}_{\lambda }(3000\,\mathring{{\rm{A}}})),\end{eqnarray} \tag{ 7 }$$

yields L_bol = 6.5 × 10⁴⁶ erg s⁻¹ and an Eddington ratio of L_bol/L_Edd = 1.8. For the reminder of the analysis we consider the bolometric correction from Equation (4) to facilitate direct comparison with relevant literature (e.g., Shen et al. 2019; Schindler et al. 2020).

In Figure 5 we plot black hole mass versus bolometric luminosity for P172+18 as well as other z > 6 and lower-redshift quasars from the literature. As for Figure 4, the low-redshift quasar sample is taken from the SDSS DR7 quasar catalog. Here, we select objects with redshift 0.35 < z < 2.25, i.e., for which the Mg ii emission line falls within the observed wavelength range, and with valid values of FWHM(Mg ii) and L_λ (3000), necessary to estimate the black hole masses and bolometric luminosities. This results in 85,504 SDSS quasars, out of which 5769 are classified as radio-loud (red contours in Figure 5). We compiled the z > 6 quasar sample from the following studies: Willott et al. (2010b), De Rosa et al. (2011), Wu et al. (2015), Mazzucchelli et al. (2017), Shen et al. (2019), Pons et al. (2019), Reed et al. (2019), Matsuoka et al. (2019b), Onoue et al. (2019, 2020), and Yang et al. (2020). We recalculate the black hole masses and bolometric luminosities of all quasars, at both low and high redshift using Equations (4) and (5). The two high-redshift radio quasars for which these measurements are available from the literature (J1427+3312 and J1429+5447, both with near-infrared spectra presented by Shen et al. 2019), show black hole masses and bolometric luminosities consistent with radio-loud quasars at lower redshift, and with the general quasar population at z > 6. The black hole of P172+18 is accreting matter at a rate consistent with super-Eddington accretion, and it is found among the fastest accreting quasars at both z ∼ 1 and z ≳ 6.

Zoom In Zoom Out Reset image size

Near-zones are regions around quasars where the surrounding intergalactic gas has been ionized by the quasar's UV radiation, and they are observed as regions of enhanced transmitted flux close to the quasar in their rest-frame UV spectra. The near-zone sizes of quasars provide constraints on quasar emission properties and on the state of their surrounding IGM (e.g., Fan et al. 2006; Eilers et al. 2017, 2018). The radii of near-zones (R_NZ) depend on the rate of ionizing flux from the central source, on the quasar's lifetime, and on the ionized fraction of the IGM (e.g., Fan et al. 2006; Davies et al. 2019). In practice, R_NZ is measured from the rest-frame UV spectrum (smoothed to a resolution of 20 Å) and taken to be the distance from the quasar at which the transmitted continuum-normalized flux drops below 10%. Here, we obtain the transmitted flux by dividing the observed spectrum of P172+18 by a model of the intrinsic continuum emission obtained with a principal components analysis method (see Davies et al. 2018; Eilers et al. 2020, for details of the method). In order to take into account the dependence on the quasar's luminosity, we also calculate the corrected near-zone radius (R_NZ,corr), following the scaling relation presented by Eilers et al. (2017):

$$\begin{eqnarray}&&{R}_{\mathrm{NZ},\mathrm{corr}}={R}_{\mathrm{NZ}}\times {10}^{0.4(27+{M}_{1450})/2.35}\end{eqnarray} \tag{ 8 }$$

where M₁₄₅₀ is the absolute magnitude at rest-frame 1450 Å. We report both R_NZ and R_NZ,corr in Table 5. The size of the near-zone of P172+18 and the corrected near-zone size are R_NZ = 3.96 ± 0.48 pMpc and R_NZ,corr = 6.31 ± 0.76 pMpc, respectively. This large near-zone is within the top quintile of the distribution of quasar near-zones at z ≳ 6 (Eilers et al. 2017). This suggests that the time during which this quasar is UV-luminous (here referred to as its lifetime) exceeds the average lifetime of the high-redshift quasar population of t_Q ∼ 10⁶ yr (Eilers et al. 2020).

The evolution of R_NZ,corr with redshift, at z > 5.5, has been investigated in the literature to constrain both the reionization history and quasar lifetimes (e.g., Carilli et al. 2010; Davies et al. 2016; Eilers et al. 2020). While Carilli et al. (2010) and Venemans et al. (2015) recover a steep decline of R_{NZ, corr} with redshift (a decrease in size by a factor of ∼6 between z = 6 and z = 7), Eilers et al. (2017) study a larger sample of ∼30 quasars at 5.8 < z < 6.6 and recover a best-fit relation in the form of ${R}_{\mathrm{NZ},\mathrm{corr}}\propto {\left(1+z\right)}^{-\gamma }$, with γ ∼ 1.44, suggesting a more moderate evolution with redshift than previous studies (a reduction in size by only ∼20% between z = 6 and z = 7). Finally, Mazzucchelli et al. (2017) recover a flatter relation (γ ∼ 1.0), utilizing measurements of R_NZ,corr up to z ∼ 7 (see also Ishimoto et al. 2020). Using hydrodynamical simulations, Chen & Gnedin (2020) obtained a shallow redshift evolution of near-zone sizes over the redshift range probed by the current quasar sample, i.e., 5.5 < z < 7 (see also Davies et al. 2020). The expected average corrected near-zone size at z = 6.8 is 〈R_NZ,corr〉 ≈ 4.2 pMpc for the redshift evolution from Eilers et al. (2017), and 〈R_NZ,corr〉 ≈ 2.2 pMpc when assuming a steeper evolution as found by Venemans et al. (2015).

Therefore, the new near-zone measurement for P172+18 is considerably larger than the expected average size at this redshift. However, if the quasar was more luminous in the recent past and its activity is currently in a receding phase (see Section 4.1 for tentative evidence of a decrease in the quasar's radio luminosity), the large near-zone size could be explained by a higher luminosity than what is measured at the present time.

The quasar is a point source in both the follow-up L- and S-band observations with a deconvolved size smaller than 19 × 087; see Figure 3. P172+18 is well detected in both bands with S/N > 20 and the measured flux densities are listed in Table 1. The measured L-band flux density is a factor of two fainter than what is reported in the FIRST catalog. In fact, the measured f_{1.52 GHz} = 510 ± 15 μJy would have been below the detection threshold of the FIRST survey (Becker et al. 1995). The discrepancy is significant at more than 3σ and could be the result of real quasar variability over the 20 yr (∼2.5 yr rest frame) between the two measurements; such changes have been reported in similar timescales (e.g., Nyland et al. 2020). However, given that the source is at the faint limit of the FIRST survey, we cannot rule out that the variation is simply due to noise fluctuations in the FIRST data. Unfortunately, we are not able to test the variability hypothesis given that no other measurements of the quasar are available at a similar epoch to the FIRST observation. For the remainder of the analysis we will consider the follow-up VLA measurements as the true fluxes. Assuming that the radio observations follow a power-law spectral energy distribution (f_ν ∝ ν^α ), the L- and S-band flux densities correspond to a steep power-law radio slope of ${\alpha }_{S}^{L}$ = −1.31 ± 0.08. This is steeper than α = −0.75, which is usually assumed in high-redshift quasar studies when only one radio band is available (e.g., Wang et al. 2007; Momjian et al. 2014; Bañados et al. 2015).

4.1.1. Radio-loudness

To estimate the radio-loudness of P172+18 we obtain the rest-frame 5 GHz emission by extrapolating the radio emission using the measured spectral index ${\alpha }_{S}^{L}=-1.31$ and the 2500 Å and 4400 Å emission using the power-law fit to the near-infrared spectrum (α_ν,UV = −0.48) obtained in Section 3. This results in radio-loudness parameters of R₂₅₀₀ = 91 ± 9 and R₄₄₀₀ = 70 ± 7, classifying P172+18 as a radio-loud quasar. The quasar radio properties are summarized in Table 5.

We note that the data from very long baseline interferometry (VLBI) presented by Momjian et al. (2021) imply a steeper spectral index at frequencies higher than 3 GHz (see Figure 6). The quasar is not detected in the TIFR GMRT Sky Survey (TGSS; Intema et al. 2017) at 147.5 MHz. We downloaded the TGSS image and determined a 3σ upper limit of 8.5 mJy (see Table 1 and Figure 6). This implies that the slope of the radio spectrum should flatten or have a turnover between 147.5 MHz and 1.52 GHz. If the turnover occurs at a frequency higher than rest-frame 5 GHz, the rest-frame 5 GHz luminosity (and therefore radio-loudness) would be smaller than our fiducial value assuming α = −1.31. In the extreme case that the turnover happened exactly at the frequency of our L-band observations, the source would still be classified as radio-loud (i.e., R₂₅₀₀ > 10) as long as α < 1.24 (see Figure 6). Deep radio observations at frequencies <1 GHz are needed to precisely determine the rest-frame 5 GHz luminosity and the shape of the radio spectrum.

Zoom In Zoom Out Reset image size

The radio companion is detected with S/N > 20 in both L- and S-band observations (see Figure 3). This object is a point source in the L-band image with a deconvolved size smaller than 16 × 069. A Gaussian fit to the S-band image results in a resolved source with a deconvolved size of 13 × 08 and position angle of 74° ± 22°. This secondary source is not detected in any of our available optical, near-infrared, and mid-infrared images. Its radio properties and optical/near-infrared limits are listed in Table 1.

The number of radio sources with a 1.4 GHz flux density >700 μJy is 59 deg⁻² and 117 deg⁻² according to the number counts of deep radio surveys from Fomalont et al. (2006) and Bondi et al. (2008), respectively. This means that in an area encompassing the quasar and the second radio source (π × 231²) only 0.007 and 0.015 sources like the companion are expected using the number counts from Fomalont et al. (2006) and Bondi et al. (2008), respectively. The <2% likelihood of chance superposition raises the possibility that this radio source and the quasar could be associated.

This companion radio source is (slightly) brighter than the quasar in both the L- and S-band follow-up observations. However, it was not detected in the FIRST survey carried out in 1999 (see Table 1 and Figure 3). This second source could not be a hot spot of the radio jet expanding for the last 20 yr: at the redshift of the quasar, the projected separation of the two sources is about 120 proper kpc, a distance that would take light about 400,000 yr to travel.

Another possibility is that this second source is an obscured, radio-AGN companion. There are a few examples of associated dust-obscured, star-forming companion galaxies to quasars at z > 6 (e.g., Decarli et al. 2017; Neeleman et al. 2019). A couple of them have tentative X-ray detections, which make them obscured AGN candidates (e.g., Connor et al. 2019; Vito et al. 2019). This possibility is tempting, because two associated radio-loud AGNs would point to an overdense environment in the early universe and provide constraints on AGN clustering. Nevertheless, with the available shallow optical and near-infrared data we are not able to rule out that the second radio source lies at a different redshift than that of the quasar. More follow-up observations are required to firmly establish the nature and redshift of the source.