Subaru SHELLQs: Low-Luminosity High-z Quasars

• The project SHELLQs delivers the most complete census of faint high-z quasars, refining the quasar luminosity function and constraining early SMBH growth models.
• The survey employs deep optical and near-infrared imaging with Subaru’s HSC and Bayesian candidate selection to achieve high spectroscopic completeness for z > 6 quasars.
• SHELLQs findings reveal sub-Eddington black hole masses and coeval host galaxy properties, challenging the quasar contribution to cosmic reionization.

The Subaru High-z Exploration of Low-Luminosity Quasars (SHELLQs) project is a wide-area, deep optical and near-infrared survey focused on the identification and characterization of faint quasars at high redshift ($z \gtrsim 6$), utilizing data from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP). SHELLQs has redefined the demographic and physical understanding of the early quasars that dominate the faint end of the quasar luminosity function (QLF), offering new constraints on the growth of supermassive black holes (SMBHs), the co-evolution of SMBHs and their host galaxies, the nature of AGN feedback, and the contribution of active galactic nuclei (AGN) to cosmic reionization.

1. Survey Strategy, Candidate Selection, and Samples

SHELLQs leverages the 1.77 deg$^2$ field-of-view and sub-arcsecond seeing of Subaru’s HSC to execute a multi-band optical survey covering $>1000$ deg$^2$ to depths of $z_\text{AB} \sim 25.1$ mag and $y_\text{AB} \sim 24.4$ mag. Candidates are selected using a Bayesian probabilistic algorithm, combining color selection (Lyman break: $i$-dropouts for $z\sim6$; $z$-dropouts for $z\sim7$), morphology (point-source cuts: $0.7 < \mu/\mu_\text{PSF} < 1.2$ in $z$ band), and near-infrared photometry to discriminate against brown dwarfs and low-$z$ interlopers (Matsuoka et al., 2016, Matsuoka et al., 2021). Spectroscopic completeness is high for $z_\text{AB}<24.5$ candidates, with outlier and artifact rejection enhanced by forced-photometry and visual inspection.

To date, SHELLQs has confirmed 139 broad-line quasars, 23 narrow-line/type-II quasar candidates, and numerous low-luminosity AGN, galaxies, and contaminants (brown dwarfs, [O III] emitters) at $5.7 < z < 7.1$ (Matsuoka et al., 28 Aug 2025). The project uniquely pushes down to $M_{1450} \sim -21$ mag, three to four magnitudes fainter than previous large-area surveys.

2. Quasar Luminosity Function and Reionization Contribution

SHELLQs measurements of the QLF at $z\sim6-7$ are based on close to 150 spectroscopically confirmed quasars, delivering the largest high-$z$ low-luminosity quasar sample to date (Matsuoka et al., 2018, Matsuoka et al., 2023). The QLF is well fit by a double power law: $$\Phi(M_{1450}) = \frac{\Phi^*}{10^{0.4(\alpha+1)(M_{1450} - M_{1450}^*)} + 10^{0.4(\beta+1)(M_{1450} - M_{1450}^*)}}$$ with a break at $M_{1450}^* \sim -24.9$ ($z=6$) and $M_{1450}^* \sim -25.6$ ($z=7$), bright-end slope $\beta \sim -2.7$ to $-3.3$, and notably flat faint-end slope $\alpha \approx -1.2$. The QLF displays pronounced flattening toward the faint end, confirmed over $-30 < M_{1450} < -22$ mag at $z\sim6$ and $-28 < M_{1450} < -23$ mag at $z=7$ (Matsuoka et al., 2023). SHELLQs data rule out a steep, upturning faint-end slope, contrary to earlier extrapolations from luminous samples.

The acceleration in the decline of quasar number density at $z>6$ is best parameterized by $\rho \propto 10^{-kz}$ with $k \sim -0.78$ ($6

The ionizing photon density from quasars, as inferred from the SHELLQs QLF, is

$$\dot{n}_{\rm ion} \sim 10^{48.8}~\mathrm{s}^{-1}\,\mathrm{Mpc}^{-3}~(z=6),~10^{48.2}~\mathrm{s}^{-1}\,\mathrm{Mpc}^{-3}~(z=7)$$

well below the critical rate required to keep the IGM ionized ($<10\%$ at $z=6$, $<1\%$ at $z=7$). Thus, luminous and faint quasars are not dominant reionization sources; the role falls instead to star-forming galaxies (Matsuoka et al., 2018, Matsuoka et al., 2023).

3. Black Hole Properties and Mass Determination

SHELLQs systematically measures SMBH masses for low-luminosity high-$z$ quasars using multiple techniques. Near-infrared spectroscopy (e.g., Mg II $\lambda2798$) provides virial mass estimates for a subset ($M_{\mathrm{BH}}=10^{7.6}$–$10^{9.3}~M_\odot$, often at sub-Eddington ratios $L_{\mathrm{bol}}/L_{\mathrm{Edd}}=0.16$–1.1) (Onoue et al., 2019). For the large photometric sample, a novel "spectral counterpart" method matches high-z rest-UV spectra (1200–1400 Å) to low-z SDSS templates with known C IV FWHM, allowing $M_{\mathrm{BH}}$ estimation: $$\log \left( \frac{M_\mathrm{BH}}{M_\odot} \right) = 0.66 + 0.53 \log \left( \frac{\lambda L_\lambda}{10^{44}~\mathrm{erg\,s}^{-1}} \right) + 2\log\left( \frac{\mathrm{FWHM}}{\mathrm{km\,s}^{-1}} \right)$$ (Takahashi et al., 2023). The $z\sim6$ SHELLQs sample peaks at $M_{\mathrm{BH}} \sim 10^{8}$–$10^{8.6}~M_\odot$, with the majority accreting sub-Eddington, in contrast to the near-Eddington and hyper-massive ($>10^{9}~M_\odot$) SMBHs seen in luminous SDSS quasars. This "normal" population is compatible with standard seed and accretion models if super-Eddington growth or more massive seeds are invoked only in the most extreme sources.

4. Host Galaxies, Dark Matter Halos, and Black Hole–Galaxy Coevolution

Spatially resolved ALMA observations provide direct measurements of [C II] 158 μm emission and FIR continuum, constraining SFRs, ISM physical conditions, and dynamical masses of quasar hosts (Izumi et al., 2019, Izumi et al., 2018, Izumi et al., 2021, Izumi et al., 2021). SHELLQs hosts are massive, gas-rich galaxies, with [C II] and FIR luminosities $L_{\rm [CII]} = (2.4$–$9.5)\times 10^8~L_\odot$, SFRs of 16–63 $M_\odot$ yr$^{-1}$ (up to $>100\,M_\odot$ yr$^{-1}$ in some cases), and $M_{\rm dyn}\sim 10^{10}$–$10^{11}\,M_\odot$ (Izumi et al., 2019, Izumi et al., 2018).

The SMBH-to-host dynamical mass ratios for low-luminosity SHELLQs quasars are on or below the local $M_{\mathrm{BH}}/M_{\mathrm{bulge}}$ relation, in contrast to the "overmassive" SMBHs of optically luminous high-z quasars. This observational result supports early, rapid, but coeval growth of SMBHs and their hosts, requiring prior starburst phases to amass stellar mass (Izumi et al., 2019, Izumi et al., 2021).

Clustering analysis using the spatial auto-correlation of 107 spectroscopically confirmed $z\approx6$ quasars establishes the typical dark matter halo mass: $$M_\mathrm{DMH} = 5.0_{-4.0}^{+7.4} \times 10^{12} \, h^{-1} M_\odot$$ with the bias parameter $b=20.8\pm8.7$ (Arita et al., 2023). No quasars are found in halos above $10^{13}~h^{-1}M_\odot$; the characteristic halo mass for quasar activation is nearly constant ($10^{12.5}~h^{-1}M_\odot$) from $z\sim2$ to $z\sim6$, indicating quasar ignition is tied to a distinct phase in the evolution of massive halos. The inferred duty cycle at $z\sim6$ is low ($f_\mathrm{duty}\approx 0.02$), dictating that observable quasar phases are brief and rare at a given halo mass.

5. AGN Feedback, Outflows, and Circumgalactic Phenomena

ALMA imaging of [C II] in select SHELLQs objects at $z>6$ has directly detected spatially extended complex emission, galactic-scale molecular and atomic outflows (e.g., velocities of $630$–$780$ km s$^{-1}$, outflow rates $\gtrsim100$–$1400\,M_\odot$ yr$^{-1}$), and dynamical structures consistent with rotating bulges (Izumi et al., 2021, Izumi et al., 2021). In red quasars—interpreted as transition objects between dust-obscured starbursts and type-I quasars—such outflows coexist with vigorous star formation ($>100\,M_\odot$ yr$^{-1}$). However, outflow energetics (kinetic power and momentum) are lower than predicted by energy-driven feedback models, indicating AGN feedback is not always sufficient to quench host star formation.

The detection rate and properties of Ly$\alpha$ haloes in faint quasars at $z\sim4$–6 show exponential surface brightness profiles with flattening at $r\lesssim5$ kpc and systematic faintness and compactness relative to bright quasars. Ly$\alpha$ halo luminosities correlate with both quasar UV and Ly$\alpha$ peak luminosities, with powering attributed primarily to recombination in optically thin, dense CGM, and secondary contributions from optically thick clumps and resonant scattering, especially in central regions (Hoshi et al., 8 Apr 2025).

6. Populations of Obscured, Narrow-Line, and Red Quasars

SHELLQs identified and characterized type-II/narrow-line AGN candidates at $z>6$—objects with faint continua, extremely luminous and narrow Ly$\alpha$ ($L_{\mathrm{Ly}\alpha} > 10^{44}$ erg s$^{-1}$, FWHM$<230$–$500$ km s$^{-1}$), and rest-frame equivalent widths $>300$ Å (Onoue et al., 2021, Iwasawa et al., 7 May 2025). Rest-optical JWST/NIRSpec spectroscopy reveals weak, partially obscured broad H$\alpha$ emission and strong [O III]$\lambda$5007 lines. Chandra X-ray observations fail to detect these objects ($L_X < 2\times10^{44}$ erg s$^{-1}$ upper limits), whereas their emission line strengths suggest $L_X\sim10^{45}$ erg s$^{-1}$ if unobscured. This is consistent with Compton-thick, orientation-dependent obscuration or burial of the nuclei in dense ISM or super-Eddington accretion disks. These hidden AGN are analogous to JWST-"Little Red Dots" and likely represent a large, previously uncounted population of SMBHs in the reionization era.

A small fraction of "red" quasars with significant dust reddening ($E(B-V)>0.1$) have also been confirmed, all at lower luminosities, providing key targets to study feedback "blowout" phases (Kato et al., 2020).

7. Legacy, Impact, and Connection to Early Structure Formation

SHELLQs has established the most complete, least-biased census of $z>6$ low-luminosity quasars, providing definitive measurements of the faint-end QLF, host galaxy and dark halo properties, black hole masses and accretion states, and ionizing photon contribution. Its sample—uniformly selected and publicly available—is foundational for studies of SMBH–galaxy coevolution, AGN feedback and obscuration physics, and reionization, and is the prime database for JWST, ALMA, and future facilities. SHELLQs bridges the luminosity and redshift gaps between classical broad-line SDSS quasars and JWST-selected faint AGN, revealing the full spectrum of early SMBH and galaxy growth (Matsuoka et al., 28 Aug 2025, Takahashi et al., 2023).

SHELLQs results demonstrate that black hole–host scaling relations, characteristic host halo mass, and feedback-regulated AGN lifetimes are already established by $z\sim6$–7. Quasars are found to be rare, short-lived phenomena in massive halos, consistent with self-regulated accretion and feedback. The absence of an abundant faint AGN population or dominant quasar contribution to reionization is now robustly established. The emergence of numerous obscured and transition AGN at these epochs further complicates the census, emphasizing the importance of multiwavelength and spectroscopic surveys.

---

Table: SHELLQs Quasar and Host Properties at $z \sim 6$

Quantity	Value(s)	Reference (arXiv)
QLF faint-end slope, $\alpha$	$-1.23^{+0.44}_{-0.34}$ ($z=6$), $-1.2$ ($z=7$)	(Matsuoka et al., 2018, Matsuoka et al., 2023)
Typical $M_{\rm BH}$	$\sim 10^{8}$–$10^{8.6}~M_\odot$	(Takahashi et al., 2023)
Typical $M_{\rm DMH}$	$5.0_{-4.0}^{+7.4} \times 10^{12}~h^{-1}M_\odot$	(Arita et al., 2023)
Typical host $M_{\rm dyn}$	$10^{10}$–$10^{11}~M_\odot$	(Izumi et al., 2019, Izumi et al., 2018)
Duty cycle, $f_\mathrm{duty}$	$0.019 \pm 0.008$	(Arita et al., 2023)
Ionizing photon rate, $\dot{n}_{\rm ion}$	$10^{48.8}$ s$^{-1}$ Mpc$^{-3}$ ($z=6$), $10^{48.2}$ s$^{-1}$ Mpc$^{-3}$ ($z=7$)	(Matsuoka et al., 2018, Matsuoka et al., 2023)
Host $L_{\rm [CII]}$	$(2.4$–$9.5)\times 10^8~L_\odot$	(Izumi et al., 2019)

---

SHELLQs provides a critical empirical testbed for theories of early SMBH seeding, galaxy formation, and the physical interplay between black holes and their environments at cosmic dawn.