A deep dive down the broad-line region: permitted OI, CaII and FeII emission in an AGN Little Red Dot at z=5.3

TL;DR

This study leverages JWST/NIRCam and NIRSpec data to reveal a broad-line AGN hosted by Bz5.3 at $z_{\rm spec} = 5.2907$, including the first detection of the [1304] bump at $z>5$. The analysis demonstrates a stratified BLR with multiple gas phases, evidenced by broad Balmer lines coexisting with narrower low-ionization lines such as O I and Ca II, and supported by strong UV/optical Fe II emission. Ly$\beta$ fluorescence emerges as the primary driver of the low-ionization lines, with collisional excitation contributing, while recombination and continuum fluorescence are disfavored; dust attenuation within the BLR is inferred from the suppressed $[1304]/[8446]$ photon flux ratio, yielding $A_V \sim 0.4$--$1.4$. The results provide a rare, detailed window into BLR physics at the highest accessible redshifts and highlight the need for higher-resolution follow-up to fully characterize the BLR structure and excitation in the early Universe.

Abstract

We present a spectroscopic analysis of a broad-line active galactic nucleus (AGN) selected as little red dot at $z = 5.3$ behind the Bullet cluster (Bz5.3), based on JWST/NIRCam and NIRSpec data. The detection of strong FeII, OI, and CaII triplet emission lines, along with the evidence of broad Balmer lines, provides strong evidence of a broad-line region (BLR) and an accreting supermassive black hole. Notably, we report the first detection of the $\lambda1304$ bump (i.e., blend of OI$λ$1304 and SiII) at high redshift, a feature commonly seen in local AGNs but not yet reported in the early Universe. The OI$λ$1304/$\lambda8446$ photon ratio provides an independent measurement of dust attenuation in galaxies. In Bz5.3, this ratio is highly suppressed (0.1--0.3), implying significant internal dust extinction, with estimated dust attenuation $A_V \sim 0.4$--$1.0$. We identify Ly$β$ fluorescence as the dominant excitation mechanism of the low-ionization lines, with additional contributions from collisional excitation. High OI$λ$8446 equivalent width and weak OI$λ$7774 support this interpretation. The detection of iron emission, whether from broad permitted or narrow forbidden lines, supports the presence of a stratified BLR, as also recently proposed in local LRDs. Photoionization modeling of OI$λ$8446 and CaII further suggests the coexistence of multiple gas phases with distinct densities and ionization states, highlighting the complexity of the BLR. Bz5.3 thus offers a rare window into early AGN activity and BLR physics at early times.

Figures (7)

• Figure 1: HST/ACS and NIRCam photometry, and NIRSpec spectra of Bz5.3. Top panels: 2D and 1D NIRSpec spectra. Fluxes are not corrected for magnification. Lines of interest for this work are highlighted by colored vertical bars. Other detected emission lines are marked with dashed gray vertical lines. The y-axis scale of the 1D spectrum is logarithmic. For comparison the spectrum of A2744-45924 at $z=4.47$ is shown in light orange. Bottom panels: NIRCam cut-out photometry in different filters, specified at the top left corner of each panel, and RGB image made combining the F444W, F356W, and F277W for red, F150W and F200W for green, and F090W and F115W filters for blue. Cutouts images are 2"$\times$2" in size. In each panel, there is an inset zooming-in on the center region to show the flux centroid of the filter (red cross), compared to the flux centroid in F444W (black cross). The PSF size of each NIRCam filter is shown as a gray circle in the bottom right corner of each panel (or zoom-in inset).
• Figure 2: Zoom-in of the 2D and 1D spectra of Bz5.3. The observed spectrum is shown in black, and the best-fitting model is shown in red. Residuals are shown at the bottom of each panel, where the shaded gray area represents the 1$\sigma$ noise level. Left panel: zoom on the [1304] bump. The vertical pink line is at the laboratory rest wavelength of the [1304] emission line. The vertical dashed black line marks the Ly$\alpha$ line. Right panel: zoom-in on the [8446] emission line and on the triplet. Vertical lines are set at laboratory rest wavelength for these emission lines. The scale of the y-axis is linear.
• Figure 3: Zoom-in of the 2D and 1D spectra of Bz5.3. The observed spectrum is shown in black, while the best-fitting model is shown in blue. Residuals are shown at the bottom of each panel, where the shaded dark gray area represents the 1$\sigma$ noise level. Top panel: zoom on the emission in the optical. The vertical lines are at the laboratory rest wavelength of the reported emission line. Bottom panel: zoom on the complex in the UV and emission line. 2 and 3$\sigma$ noise levels are reported in the residual as shaded lighter gray area.
• Figure 4: Results of modeling Bz5.3 with Galfit. Left panels: data and results assuming a single point source as model for each filter. The white cross indicates the centroid of the model. Right panels: data and results assuming as model a combination of a point source and a Sérsic profile for each filter with $\lambda<4~\mu m$, a single point source otherwise.
• Figure 5: Zoom-in of the 2D and 1D spectra of Bz5.3 on the complex in the UV and emission line. The observed spectrum is shown in black, while the best-fitting model is shown in blue. Residuals are shown at the bottom of each panel, where the shaded dark gray area represents the 1$\sigma$ noise level. 2 and 3$\sigma$ noise levels are reported in the residual as shaded lighter gray area.