JADES: A large population of obscured, narrow line AGN at high redshift

TL;DR

JWST/NIRSpec observations from the deepest JADES fields enable the identification of a large population of obscured, narrow-line (type-2) AGN up to z~9. The study employs a multi-diagnostic approach combining optical (N2-BPT, S2-VO87, He II) and UV (C3He2, C43, NeIV, NeV, NV) line ratios, anchored by CLOUDY photoionization models, to robustly distinguish AGN from star-forming galaxies in a low-metallicity, high-ionization regime. They find ~20% of galaxies host type-2 AGN with bolometric luminosities spanning 10^{41.3}–10^{44.9} erg s^{-1} and host masses around 10^{7.2}–10^{9.3} M⊙, with SFRs consistent with the star-forming main sequence; AGN contribute notably to the UV luminosity function at z~4–6 (8–30% depending on luminosity). A striking finding is a robust z~9.43 type-2 AGN, illustrating JWST’s power to probe SMBH growth in the early universe and implying an appreciable population of faint, X-ray-weak AGN that influence galaxy evolution and reionization models.

Abstract

We present the identification of 42 narrow-line active galactic nuclei (type-2 AGN) candidates in the two deepest observations of the JADES spectroscopic survey with JWST/NIRSpec. The spectral coverage and the depth of our observations allow us to select narrow-line AGNs based on both rest-frame optical and UV emission lines up to z=10. Due to the metallicity decrease of galaxies, at $z>3$ the standard optical diagnostic diagrams (N2-BPT or S2-VO87) become unable to distinguish many AGN from other sources of photoionisation. Therefore, we also use high ionisation lines, such as HeII$λ$4686, HeII$λ$1640, NeIV$λ$2422, NeV$λ$3420, and NV$λ$1240, also in combination with other UV transitions, to trace the presence of AGN. Out of a parent sample of 209 galaxies, we identify 42 type-2 AGN (although 10 of them are tentative), giving a fraction of galaxies in JADES hosting type-2 AGN of about $20\pm3$\%, which does not evolve significantly in the redshift range between 2 and 10. The selected type-2 AGN have estimated bolometric luminosities of $10^{41.3-44.9}$ erg s$^{-1}$ and host-galaxy stellar masses of $10^{7.2-9.3}$ M$_{\odot}$. The star formation rates of the selected AGN host galaxies are consistent with those of the star-forming main sequence. The AGN host galaxies at z=4-6 contribute $\sim$8-30 \% to the UV luminosity function, slightly increasing with UV luminosity.

Figures (15)

• Figure 1: Example of low-resolution PRISM/CLEAR spectrum of galaxy (JADES-NS-GS-00022251) at z = 5.804 included in our parent sample. The spectrum is shown in units of F$_\lambda$ normalized to the [O iii] $\lambda$5008 peak. We highlight major emission lines used in our analysis.
• Figure 2: List of AGN from JADES HST Deep selected in this work. Columns: ID, redshift, detection method $^{+}$, stellar mass, SFR and UV magnitude from BEAGLE, AGN bolometric luminosity and any other additional notes. We note that for bolometric luminosity the uncertainties are dominated by the systematical uncertainties (0.3-0.5 dex).
• Figure 3: Typical line ratio diagnostic diagrams used to select AGN. N2-R3 BPT ([N ii]/H$\alpha$ vs [O iii]/H$\beta$; top row) and S2-VO87 ([S ii]/H$\alpha$ vs [O iii]/H$\beta$; bottom row). The ionisation models of Feltre16Gutkin16 (left) and Nakajima22 (right) are reported as yellow (SF) and blue (AGN) points (see § \ref{['sec:cloudy']}). The objects from this work are plotted as blue squares. The objects selected as AGN based in these diagrams are highlighted by red circles. We also plot our new demarcation lines as green dashed. The black dashed lines show the star forming versus AGN demarcation lines from Kewley01 and Kauffmann03. For comparison, we plot SDSS galaxies shown as a grey contour plot. The magenta and cyan squares show a stacked spectrum for AGN and star-forming galaxies (see § \ref{['sec:stacking']}). We highlight type-1 AGN from Harikane23 and Maiolino23JADES as green squares and diamonds. The red star shows the X-ray selected AGN in our sample.
• Figure 4: The He2-N2 (He ii /H$\beta$ vs [N ii]/H$\alpha$) diagram for the sample of JADES galaxies for our sample, plotted as blue squares. The left and right plots show ionisation models from Feltre16Gutkin16 and Nakajima22, respectively, as yellow and light blue points (see § \ref{['sec:cloudy']}). The magenta and cyan squares show a stacked spectrum for AGN and star-forming galaxies (see § \ref{['sec:stacking']}). The black dashed line indicates a demarcation line between star-forming and AGN galaxies by Shirazi12. The blue and black contours show the star-forming galaxies and AGN from SDSS, respectively. We highlight the selected AGN in this diagram with a red circle.
• Figure 5: C iii]/He ii$\lambda$1640 vs C iv/C iii] (CHe2-C43) diagnostic diagram. We plot our sample as blue squares. The left and right plots show ionisation models from Feltre16Gutkin16 and Nakajima22, respectively, as yellow and light blue points (see § \ref{['sec:cloudy']}). The green diamonds show He ii$\lambda$1640 detection from the VANDELS survey Saxena20, and all C iv detections from Mascia23 as grey colour with squares and circles representing SF and AGN, respectively. The green squares show local analogues of high-redshift galaxies Mingozzi23. We also show the black dashed and dotted demarcation lines between AGN, star-forming galaxies and composite line ratios from Hirschmann22. Overall, we select five AGN on this diagram and we highlight these with a red circle. The magenta and cyan squares show a stacked spectrum for AGN and star-forming galaxies (see § \ref{['sec:stacking']}).