Pōniuā‘ena: A Luminous $z=7.5$ Quasar Hosting a 1.5 Billion Solar Mass Black Hole

Luminous reionization-era quasars ($z>6.5$) provide unique probes of supermassive black hole (SMBH) growth, massive galaxy formation, and intergalactic medium (IGM) evolution in the first billion years of the Universe’s history. However, efforts to find such objects have proven to be difficult because of a combination of the declining spatial density of quasars at high redshift, the limited sky coverage of near-infrared (NIR) photometry, and the low efficiency of spectroscopic follow-up observations.

During the past few years, high-redshift quasar searches using newly available wide-area optical and IR surveys have resulted in a sixfold increase in the number of known quasars at $z>6.5$: 47 luminous quasars at $z>6.5$ have been discovered (e.g., Fan et al., 2019; Venemans et al., 2013, 2015; Wang et al., 2019; Mazzucchelli et al., 2017; Reed et al., 2019; Matsuoka et al., 2019a; Yang et al., 2019), although among them only six are at $z\geq 7$ (Mortlock et al., 2011; Wang et al., 2018; Matsuoka et al., 2019a, b; Yang et al., 2019) and one at $z>7.1$ (Bañados et al., 2018). These discoveries show that 800 million solar-mass black holes exist already at $z=7.5$ (Bañados et al., 2018) and that the IGM is significantly neutral at $z\gtrsim 7$ (e.g. Bañados et al., 2018; Davies et al., 2018b; Greig et al., 2017, 2019; Wang et al., 2020). However, both early SMBH growth history and IGM neutral fraction evolution at $z>7$ are still poorly constrained because of the small sample size. For statistical analysis, more $z\sim 7-8$ quasars are necessary to investigate the IGM, SMBH masses, and quasar host galaxies at this critical epoch.

In this paper, we report the discovery of a new quasar J100758.264+211529.207 (“Pōniuā‘ena” in the Hawaiian language, hereinafter J1007+2115) at $z=7.5149$. This object is only the second quasar known at $z\sim 7.5$, close to the mid-point redshift of reionization (Planck Collaboration et al., 2018). Its discovery enables new measurements of a quasar Ly$\alpha$ damping wing and provides new constraints on the earliest SMBH growth. In this paper, we adopt a $\Lambda$CDM cosmology with parameters $\Omega_{\Lambda}=0.7$, $\Omega_{m}=0.3$, and $h=0.685$. Photometric data are reported on the AB system after applying a Galactic extinction correction (Schlegel et al., 1998; Schlafly & Finkbeiner, 2011).

In this section we describe the selection method that led to the discovery of J1007+2115 and the spectroscopic observations. This quasar was selected based on the same photometric dataset used for our previous $z\sim 6.5-7$ quasar surveys (Wang et al., 2018, 2019; Yang et al., 2019) but with selection criteria focused on a higher redshift range.

The central black hole mass of the quasar can be estimated based on its luminosity and the FWHM of the Mg II line. We fit the near-IR spectrum with a pseudo-continuum, including a power-law continuum, Fe II template (Vestergaard & Wilkes, 2001; Tsuzuki et al., 2006), and Balmer continuum (De Rosa et al., 2014). Gaussian fits of the C IV and Mg II emission lines are performed on the continuum-subtracted spectrum. A two-component Gaussian profile is used. The uncertainty is estimated using 100 mock spectra created by randomly adding Gaussian noise at each pixel with its scale equal to the spectral error at that pixel (e.g. Shen et al., 2019; Wang et al., 2020). All uncertainties are then estimated based on the 16th and 84th percentile of the distribution. The best-fit pseudo continuum and the line fitting of C IV and Mg II are shown in Figure 1.

From the spectral fit, we find that the power-law continuum has a slope $\alpha=-1.14\pm 0.01$ ($f_{\lambda}\propto\lambda^{\alpha}$). The rest-frame 3000 Å luminosity is measured to be $\lambda L_{\rm 3000}=(3.8\pm 0.2)\times 10^{46}$ erg s^-1, corresponding to a bolometric luminosity of $L_{\rm bol}=(1.9\pm 0.1)\times 10^{47}$ erg s^-1 assuming a bolometric correction factor of 5.15 (Richards et al., 2006). The apparent and absolute rest-frame 1450 Å magnitudes are derived to be $m_{\rm 1450,AB}=20.43\pm 0.07$ and $M_{\rm 1450,AB}=-26.66\pm 0.07$ from the best-fit power law continuum. The line fitting of Mg II yields an FWHM = $3247\pm 188$ km s^-1 and a Mg II-based redshift of $z_{\rm MgII}=7.494\pm 0.001$, implying a $736\pm 35$ km s^-1 blueshift relative to the [C II] line, similar to other $z\gtrsim 7$ luminous quasars (e.g., Mortlock et al., 2011; Bañados et al., 2018; Wang et al., 2020). The C IV fitting results in an FWHM of 6821$\pm$2055 km s^-1. The C IV line has a 3220$\pm$362 km s^-1 blueshift compared to the Mg II line. These measurements are summarized in Table 1.

We estimate the black mass based on the bolometric luminosity and the FWHM of the Mg II line by adopting the local empirical relation from Vestergaard & Osmer (2009). The black hole mass is derived to be $M_{\rm BH}=(1.5\pm 0.2)\times 10^{9}$ $M_{\odot}$, resulting in an Eddington ratio of $L_{\rm bol}/L_{\rm Edd}=1.1\pm 0.2$. Note that the black hole mass uncertainty estimated here does not include the systematic uncertainties of the scaling relation, which could be up to $\sim 0.5$ dex. The uncertainty of the Eddington ratio is subject to the same systematic uncertainty as the black hole mass.

Observations of previously known luminous $z\gtrsim 6.5$ quasars (e.g., Mortlock et al., 2011; Wu et al., 2015; Bañados et al., 2018) have already raised the question of how these early SMBHs grew in such a short time (e.g., Volonteri, 2012; Smith et al., 2017; Inayoshi et al., 2019), which probably requires massive seed black holes, as illustrated in Figure 2. The black hole of J1007+2115, which is twice as massive as that of the other $z=7.5$ quasar J1342+0928, further exacerbates this early SMBH growth problem. To reach the observed SMBH mass at $z=7.5$, a seed black hole with a mass of $\sim 10^{4}\,(\mathrm{or}~{}3\times 10^{5})\,M_{\rm\odot}$ would have to accrete continuously at the Eddington limit starting at $z=30\,(\mathrm{or}~{}15)$, assuming a radiative efficiency of 0.1 (see Figure 2). Under this same set of fixed assumptions about black hole growth, J1007+2115 requires the most massive seed black hole compared to any other known quasar. This is consistent with the direct collapse black hole seed model rather than the Pop III stellar remnant seed model. Even with a massive seed black hole, Eddington accretion with a high duty cycle and low radiative efficiency ($\sim 0.1$) is required. A lower mass seed would imply an even higher accretion rate (e.g., hyper-Eddington accretion) or a lower radiative efficiency (Davies et al., 2019). It has been suggested that maintaining super-Eddington accretion might be possible in specific environments (Inayoshi et al., 2016; Smith et al., 2017), but whether or not this mode of rapid black hole growth is sustainable remains an important open question.

At $z>7$, the damping wing profile, detectable as absorption redward of the Ly$\alpha$ emission line caused by the highly neutral IGM, is one of the most promising tracers of the IGM neutral fraction. J1007+2115 provides us with a new sightline to estimate the IGM neutral fraction through damping wing analysis at a time deep into the reionization epoch.

To estimate the IGM neutral fraction through damping wing analysis, we follow the procedures described in Davies et al. (2018a, b), which has also been used to analyze the spectra of three other luminous $z\gtrsim 7$ quasars (Davies et al., 2018a; Wang et al., 2020). Briefly, we first model the quasar intrinsic continuum around the Ly$\alpha$ region using the principal component analysis (PCA) approach in Davies et al. (2018b). This approach predicts the intrinsic blue-side quasar spectrum (rest-frame 1175–1280 Å) from the red-side spectrum (1280–2850 Å) using a training sample of $\sim$13,000 quasar spectra from the SDSS/BOSS quasar catalog. We then apply the method from Davies et al. (2018a) to quantify the damping wing strength and estimate the volume-averaged neutral hydrogen fraction, $\langle x\rm_{HI}\rangle$. This method models the quasar transmission spectrum with a multi-scale hybrid model, which is a combination of the density, velocity, and temperature fields, large-scale semi-numerical reionization simulations around massive quasar-hosting halos (Davies & Furlanetto in prep), and one-dimensional radiative transfer of ionizing photons emitted by the quasar (Davies et al., 2016). We construct realistic forward-modeled representations of quasar transmission spectra, accounting for the covariant intrinsic quasar continuum uncertainty. We then perform Bayesian parameter inference on the mock spectra to recover the joint posterior probability distribution functions (PDF) of $\langle x\rm_{HI}\rangle$ and $\log t_{\mathrm{Q}}$ from the observed spectrum. In the Bayesian inference, the likelihood is computed from maximum pseudo-likelihood model parameters and the pseudo-likelihood is defined as the product of individual flux PDFs of 500 km/s binned pixels, equivalent to the likelihood function of the binned transmission spectrum in the absence of correlations between pixels (see more details in Davies et al. 2018a).

To measure $\langle x\rm_{HI}\rangle$, we set a broad $t_{\rm Q}$ range of $10^{3}\,{\rm yr}<t_{\rm Q}<10^{8}\,{\rm yr}$ with a flat log-uniform prior, and compute the posterior PDF of $\langle x\rm_{HI}\rangle$ by marginalizing over quasar lifetime. As shown in Figure 4, from the posterior PDF, we can estimate $\langle x\rm_{HI}\rangle$ and its 68% confidence interval as $\langle x\rm_{HI}\rangle=0.39^{+0.22}_{-0.13}$, which is consistent with the maximum pseudo-likelihood model parameters shown in Figure 3. To avoid possible contamination from any intervening damped Ly$\alpha$ absorber, we search for associated metal absorption. No such absorption has been found in our current spectrum. We conclude that the neutral IGM should be responsible for the damping wing features of J1007+2115. If any potential damped Ly$\alpha$ absorber plays a role in generating the damping wing feature, the IGM neutral fraction will be even lower.

The detection of damping wing signatures in two $z>7$ quasar spectra has previously provided strong evidence for a significantly neutral Universe at $z\gtrsim 7$ (e.g., Mortlock et al., 2011; Bañados et al., 2018; Davies et al., 2018a). Specifically, neutral gas fractions of $\langle x\rm_{HI}\rangle\sim 0.48$ at $z=7.09$ and $\langle x\rm_{HI}\rangle\sim 0.60$ at $z=7.54$ have been reported (Davies et al., 2018a). Recent analysis of the damping wing feature of the quasar J0252–0503 at $z=7.0$ (Wang et al., 2020) also suggests a highly neutral IGM with $\langle x\rm_{HI}\rangle=0.7$. All of these measurements are based on the same methodology used in this work. We compare our result with these estimates, as shown in Figure 4. It is evident that the damping wing absorption is much weaker in J1007+2115 compared to that in the other three $z>7$ quasars. At the resonant Ly$\alpha$ wavelength, the observed spectrum of J1007+2115 does not deviate from the blue-side prediction based on red-side PCA reconstruction. This result is to be compared with J0252–0503 (Wang et al., 2020) where the observed spectrum is $\sim 40$% lower than the prediction without damping wing absorption. The $\langle x\rm_{HI}\rangle$ estimated from J1007+2115 at $z=7.54$ is lower than the measurements from all of the other three sightlines. Studies of the Ly$\alpha$ emission from $z>6$ galaxies have suggested neutral fractions of $\langle x\rm_{HI}\rangle=0.59^{+0.11}_{-0.15}$ at $z\sim 7$ and $\langle x\rm_{HI}\rangle>0.76$ at $z\sim 8$ (Mason et al., 2018, 2019). The sightline of J1007+2115 is thus a $2\sigma$ outlier, compared to the previous results. Although it is difficult to draw solid conclusions because of the large uncertainties (and the broad PDF) on the value of $\langle x\rm_{HI}\rangle$, the much weaker damping wing seen in J1007+2115’s spectrum indicates a significant scatter of the IGM neutral fraction in the redshift range $z=7.5$ to $z=7.0$, which can be interpreted as observational evidence of patchy reionization.

We report the discovery of a new quasar J1007+2115 with a [C II]-based redshift of $z=7.5149\pm 0.0004$, selected with DECaLS, PS1, UHS, and WISE photometry and observed with the Gemini, Magellan, Keck, and ALMA telescopes. The [C II] and dust continuum emission from the quasar host galaxy are well detected, and imply a SFR${}_{\mathrm{[CII]}}\sim 80-520$ $M_{\sun}$ yr^-1. It is only the second quasar known at such high redshift and thus provides a valuable new data point for early SMBH and reionization history studies.

By fitting the NIR spectrum, we derive $M_{\rm BH}=(1.5\pm 0.2)\times 10^{9}$ $M_{\odot}$ and an Eddington ratio of $L_{\rm bol}/L_{\rm Edd}=1.06\pm 0.2$ using the broad Mg II emission line. The black hole in J1007+2115 is twice as massive as that of J1342+0928 at a very similar redshift of $z=7.54$, and thus places the strongest constraint to the early SMBH growth, requiring a seed black hole with a mass of $\sim 10^{4}\,(3\times 10^{5})\,M_{\rm\odot}$ at $z=30\,(15)$. Through damping wing modeling of the quasar spectrum, we estimate the volume-averaged neutral fraction to be $\langle x\rm_{HI}\rangle=0.39^{+0.22}_{-0.13}$ at $z=7.5$. Together with three previous measurements from quasar damping wing analyses, our new result indicates a large scatter of the IGM neutral fraction from $z=7.5$ to $z=7.0$, indicative of a patchy reionization process.