The AGN nature of strong CIII emitters in the Early Universe with JWST

TL;DR

This study leverages JWST/NIRSpec prism spectroscopy to analyze $61$ robust CIII] emitters at $5 < z < 7$, combining LiMe emission-line measurements and Bagpipes SED fitting to derive physical properties and classify ionizing sources. Through broad Hα detection, UV line diagnostics, the OHNO diagram, and spectro-photometric fits, the authors find that about half of the sample shows secure AGN activity, with a median rest-frame EW(CIII]) of $22.8$ Å—significantly higher than $z \sim 3$–$4$ emitters, indicating strong redshift evolution and possible AGN-driven CIII] enhancement. Stacking analyses reveal that strong CIII] emitters occupy the AGN region in UV diagnostics, while weaker/ non-detections align with star-forming expectations, and the OHNO diagram proves less reliable at these redshifts. The paper also identifies a small subset of Little Red Dots among the CIII] emitters, highlighting the diversity of AGN activity in the early universe. Overall, the work demonstrates that AGN are common among high-$z$ CIII] emitters and underscores the need for higher-resolution follow-up to robustly separate AGN and star-formation contributions and to refine high-redshift diagnostic tools.

Abstract

The semi-forbidden CIII] $λλ$1907,1909 doublet is a key tracer of high-ionization emission in the early universe. We present a study of CIII] emission in galaxies at z=5-7, using publicly available JWST/NIRSpec prism data from programs including CEERS, JADES, RUBIES and CAPERS. We built a sample of 61 CIII]-emitting galaxies, and we classified them as star-forming or active galactic nuclei (AGN) host galaxies using (1) rest-frame UV and optical emission-line diagnostic diagrams, and (2) the presence/absence of broad Balmer emission lines. The UV diagnostics are based on the combination of the rest-frame equivalent width (EW) of CIII] versus CIII]/HeII $λ$1640, and the EW of CIV versus CIV/HeII $λ$1640. For optical diagnostics, we employ the OHNO diagram, which combines [OIII] $λ$5007, H$β$, [NeIII] $λ$3869, and [OII] $λλ$3727,3729- and we find it has a low efficiency on separating AGN from SFG. We find that half of the sources in our sample (29 out of 61 galaxies) exhibit at least one secure indication of AGN activity while 13 are potential AGNs based on the CIII] diagnostic. Physical properties, including stellar mass and star formation rate, are derived through spectral energy distribution modeling with Bagpipes. Our analysis reveals that JWST is uncovering a population of strong CIII] emitters at high redshift (5<z<7) with a median EW of 22.8 A. This EW is higher than that of a control sample of CIII] emitters at redshift 3<z<4 with a median EW of 4.7 A. We find that for the same range of Muv, the CIII] EW increases by $\sim$0.67 dex from 3<z<4 to 5<z<7, indicating strong redshift evolution in the line's strength. Finally, we identify five sources in our sample as Little Red Dots (LRDs); while four of these have already been identified as LRD in the literature, one is presented here for the first time.

Figures (13)

• Figure 1: Rest-frame spectra of CIII]-emitters from different surveys. Both panels show the rest-frame wavelength in Ångströms on the x-axis and flux density in units of $10^{-21}$ erg s$^{-1}$ cm$^{-2}$ Å$^{-1}$ on the y-axis.
• Figure 2: Redshift distribution of our galaxy sample. The histogram shows the number of galaxies as a function of redshift between 5 and 7, with a pronounced peak at 5.25.
• Figure 3: Rest-frame CIII] equivalent width (EW; in Å) versus absolute UV magnitude ($M_{\text{UV}}$; AB mag). The blue points correspond to our sample of CIII] emitters with redshifts between 5 and 7. The red points correspond to a VANDELS sample compiled by Cunningham_2024 with redshifts between 3 and 4. Contour plots and linear regression fits are also shown.
• Figure 4: Plot of the SFR vs stellar mass of our sample compared to the star formation of field galaxies at z=5-7. Our sample is color coded by the rest frame CIII] EW value. The general parent population in the same redshift range is plotted as grey circles. The blue line represents the relationship found by Calabr__2024 and derived with SFR based on $\mathrm{H}\alpha$ for a similar spectroscopic sample.
• Figure 5: Stacked spectra for the sample of non-CIII] emitters (top panel), weak (middle panel), and strong (bottom panel) CIII] emitters. In all panels, the shaded regions represent the 1$\sigma$ uncertainty of the stacking. The dashed black vertical line indicates the position of rest-frame UV emission lines.