The Rise of Faint, Red AGN at $z>4$: A Sample of Little Red Dots in the JWST Extragalactic Legacy Fields

TL;DR

The paper tackles the challenge of identifying faint, obscured AGN at high redshift by introducing a continuum-slope method that uses shifting JWST bandpasses around the Balmer break to select LRDs across z~2–11. By applying this approach to six JWST fields, it assembles 341 LRDs with a substantial AGN component, achieving a 71% AGN-confirmation fraction for the brightest subset (F444W<26.5). X-ray analysis reveals moderate obscuration in at least two LRDs, and RUBIES NIRSpec spectroscopy shows broad emission lines consistent with AGN activity, while some sources exhibit outflow signatures via blue-shifted Balmer absorption. The redshift distribution rising toward z<8 and then dropping near z~4.5 suggests a link to inside-out galaxy growth, and LRDs appear to outnumber X-ray/UV-selected AGN by about an order of magnitude at z~5–7, underscoring a crucial obscured SMBH growth phase in the early universe.

Abstract

We present a sample of 341 "little red dots" (LRDs) spanning the redshift range $z\sim2-11$ using data from the CEERS, PRIMER, JADES, UNCOVER and NGDEEP surveys. Unlike past use of color indices to identify LRDs, we employ continuum slope fitting using shifting bandpasses to sample the same rest-frame emission blueward and redward of the Balmer break. This enables the detection of LRDs over a wider redshift range and with less contamination from galaxies with strong breaks that otherwise lack a rising red continuum. The redshift distribution of our sample increases at $z<8$ and then undergoes a rapid decline at $z\sim4.5$, which may tie the emergence of these sources to the inside-out growth that galaxies experience during this epoch. We find that LRDs are $\sim1$ dex more numerous than X-ray and UV selected AGN at z~5-7. Within our sample, we have identified the first two X-ray detected LRDs. An X-ray spectral analysis confirms that these AGN are moderately obscured with $\log\,(N_{\rm H}/{\rm cm}^{2}$) of $23.3^{+0.4}_{-1.3}$ and $22.72^{+0.13}_{-0.16}$. Our analysis reveals that reddened AGN emission dominates their rest-optical light, while the rest-UV originates from their host galaxies. We also present NIRSpec observations from the RUBIES survey of 17 LRDs that show broad emission lines consistent with AGN activity. The confirmed AGN fraction of our sample is 71\% for sources with F444W<26.5. In addition, we find three LRDs with blue-shifted Balmer absorption features in their spectra, suggesting an outflow of high-density, low-ionization gas from near the central engine of these faint, red AGN.

Figures (6)

• Figure 1: Best-fit SED of the obscured, broad-line AGN CEERS 746 from Kocevski23b. The blue and red bars denote the filters blueward and redward of the Balmer break at 3645Å (the dashed vertical line) used to determine the rest-frame UV and optical continuum slopes, respectively, of each source given its redshift.
• Figure 2: Examples of our continuum slope fits for sources at a range of redshifts. The bands used to measure the rest-frame UV and optical slopes are shown as blue and red squares, respectively. The F444W image cutouts of each source are $1^{\prime\prime}\times1^{\prime\prime}$ in size. The best-fit SEDs shown in light grey are galaxy plus QSO hybrid models (see §3 for details). $2$ upper limits are shown for bands with non-detections.
• Figure 3: Distribution of best-fit optical and UV spectral slopes, $_{\rm opt}$ and $_{\rm UV}$, measured in the CEERS, PRIMER-COSMOS, PRIMER-UDS, UNCOVER, JADES, and NGDEEP datasets for galaxies at $z>2$ detected in the F444W filter with a ${\rm SNR}>12$. The horizontal and vertical dashed lines denote our selection criteria of $_{\rm opt}>0$ and $_{\rm UV}<-0.37$ meant to select sources with red and blue colors in the rest-frame optical and UV, respectively. The horizontal dotted line denotes the $_{\rm opt}$ limit that corresponds to the color cut used in Barro23 (i.e., ${\rm F277W-F444W} > 1.5$). The vertical dot-dashed line denotes the $_{\rm UV}$ limit that corresponds to the blue color cut used by Greene_2023 to exclude brown dwarfs (i.e., ${\rm F115W-F200W} < -0.5$). The red circles are sources that satisfy both our spectral slope and size cuts (see §3 for details). Also shown are sources excluded from our primary sample due to either failing our size cut (blue squares) or being flagged as potential strong line emitters (orange circles). The smaller light blue, dark blue, and pink circles are the sample of LRDs identified in Pablo24, Barro23, and Labbe23, respectively. The light blue diamonds indicate sources in our sample with broad emission line detections.
• Figure 4: F444W magnitude versus half-light radius for galaxies in the CEERS, PRIMER-COSMOS, PRIMER-UDS, UNCOVER, JADES, and NGDEEP datasets at $z>2$ detected in the F444W filter with a ${\rm SNR}>12$. Green circles are stars and the dashed line denotes our best-fit to the stellar locus. Our magnitude-dependant size cut is shown by the dotted line, which is 1.5 times our best-fit to the magnitude-size relationship for stars in each field. Symbols are the same as in Figure \ref{['fig:slopes']}
• Figure 5: Examples of our continuum slope fits for sources removed from our sample. The bands used to measure the rest-frame UV and optical slopes are shown as blue and red squares, respectively. Panels A, B and C show examples of candidate brown dwarfs cut from the sample due to their low rest-frame UV continuum slopes. The dashed green line shows the best-fit LOWZ brown dwarf atmosphere model from Meisner21. Panels D, E, and F show examples of sources whose rest-frame optical continuum slope may be boosted due to strong line emission in one or more bands. The F444W image cutouts of each source are $1^{\prime\prime}\times1^{\prime\prime}$ in size. The best-fit SEDs shown in light grey are galaxy plus QSO hybrid models (see §3 for details).