A Comprehensive JWST/NIRSpec Census of Broad-Line Active Galactic Nuclei: Faint, Tiny, but Highly Accreting Sources in the Remote Universe

TL;DR

We quantify SMBH growth by constructing a large, uniform BLAGN census with JWST/NIRSpec over $0.8 \le z \le 7.2$, using multi-Gaussian emission-line fitting of H$\alpha$ + [N II] and H$\beta$ + [O III] to isolate broad BLR components. Black hole masses are estimated via a virial relation $M_{\mathrm{BH}} \propto L_{\mathrm{H}\alpha}^{0.55} \times \mathrm{FWHM}_{\mathrm{H}\alpha}^{2.06}$, with $L_{\mathrm{bol}}$ derived from $L_{5100}$ through a bolometric correction; the sample spans $L_{\mathrm{bol}}$ ≈ $10^{43}$–$10^{45.5}$ erg s$^{-1}$ and $M_{\mathrm{BH}}$ ≈ $10^{5}$–$10^{9}\,M_\odot$, with most BLAGNs showing $\lambda_{\mathrm{Edd}}$ between $0.1$ and $1.0$. The study finds BLAGN detection rates roughly uniform up to $z \sim 7$, with high-$z$ BLAGNs typically less massive and less luminous than local counterparts but accreting more efficiently, implying rapid early SMBH growth and bridging the observational gap between local SMBHs and distant quasars. These results inform black hole–galaxy coevolution over cosmic time and demonstrate JWST/NIRSpec's power to reveal faint, low-mass AGNs in the distant universe, while acknowledging caveats from heterogeneous depths and limited high-$z$ S/N.

Abstract

We present a sample of 252 broad-line Active Galactic Nuclei (BLAGNs), incorporating 171 newly identified sources, spanning a redshift interval from $z$ = 0.8 to 7.2. We have analyzed spectroscopic data from the NIRSpec instrument aboard the James Webb Space Telescope, using the G140H, G140M, G235H, G235M, G395H, and G395M gratings to survey N $\sim$ 80,000 galaxies for BLAGNs. Through emission-line fitting, using a sum of Gaussian models for {H$α$}, {H$β$}, [N II] $λ\lambda6548, 6584$, and [O III] $ λ\lambda4959, 5007$, we separate AGN broad-line components from narrow-line emission. We find the detection rate of BLAGNs to be relatively consistent across our redshift range. Compared to typical low-$z$ AGNs ($z$ $\lesssim$ 1), the high-$z$ BLAGNs are systematically fainter and less massive, yet they accrete more efficiently, with most showing Eddington ratios between 0.1 and 1.0. This confirms the rapid black hole growth during the early cosmic epochs. The detection of faint, low-mass BLAGNs at high redshift also helps bridge the observational gap between local supermassive black holes and remote luminous quasars, providing a more complete view of black hole-galaxy coevolution across cosmic time.

Figures (9)

• Figure 1: Gaussian fittings to detect BLAGN signatures for VALENTINO-3567-51909 ($z$ = 4.626). We show the fittings for [O iii] $\lambda\lambda$ 4959, 5007 + H$\beta$ lines (left), and H$\alpha$ + [N ii]$\lambda\lambda$ 6548, 6584 (right). The NIRSpec data and errors are shown as black and gray lines, respectively. The green and blue lines are for the narrow and broad component, respectively. The red lines are the sum of both components. Inset shows an expanded view of the wing for the broad component. See Section \ref{['sec:EmissionLineFitting']} for the fitting details.
• Figure 2: The redshift distribution of our sample of BLAGNs (blue) and the total objects searched in the DJA v4 database (red) on a log scale. The detection rates of BLAGNs is almost uniform between $z = 2$ and 7, with the highest-redshift objects out to $z = 7.2$.
• Figure 3: Black hole mass (_BH $M_{\text{BH}}$) versus bolometric luminosity ($\mathrm{L}_{\text{bol}}$) for our AGN sample. Color of points indicate redshift. Symbols indicate AGNs from this study ($\star$), Willott10$z \approx 6$) ($\blacksquare$), Ubler23$z \sim 5.5$) (•), Onoue19$6.1 \le z \le 6.7$ ($\blacktriangledown$), Shen19$z \ge 5.7$($\blacktriangle$), Lin24$4 \le z \le 5$ ($\blacktriangleright$), Greene24$z \ge 5$ ($\pentagon$), Maiolino23$4 \le z \le 11$ ($+$), Harikane23$4 \le z \le 7$ ($\times$) , and Trakhtenbrot10$z \le 4.8$ ($\blacklozenge$). Light gray points are studies from which the redshift of the objects is not provided. Contours show low-$z$ AGNs from redshifts 1 -- 2 from Sloan Digital Sky Survey (SDSS) data release 16 Wu22 and the dashed lines show the Eddington ratio ($\mathrm{\lambda}_{\text{edd}}$) of 0.01, 0.1 and 1.
• Figure 4: $\mathrm{FWHM}_{\mathrm H\alpha,\text{broad}}$ versus Log $\mathrm{L}_{\mathrm H\alpha,\text{broad}}$ for our sample of BLAGNs. The points are colored by redshift. Along with our BLAGNs, we overlay other BLAGNs found by Lin24 ($\bullet$), Greene24 ($\blacktriangle$), Maiolino23 ($\blacklozenge$), Harikane23 ($\pentagon$), Trakhtenbrot10 ($\blacktriangleleft$) , and Willott10 ($\blacktriangleright$). BLAGNs identified in this study are shown as $\star$. Similarly as Figure \ref{['fig:MbhcLumbol']}, contours show low-$z$ AGNs from redshifts 1 -- 2 Wu22.
• Figure 5: Eddington ratio ($\mathrm{\lambda}_{\text{edd}}$) vs redshift for our sample. Color of points indicates black hole mass (_BH $M_{\text{BH}}$). We also include BLAGNs found in He23 as crosses. Similarly as Figure \ref{['fig:VelvsBHaLum']}, we show the lower redshift AGNs from SDSS DR16 in different contours from Wu22.