NEXUS: A Spectroscopic Census of Broad-line AGNs and Little Red Dots at $3\lesssim z\lesssim 6$

Ming-Yang Zhuang, Junyao Li, Yue Shen, Xiaojing Lin, Alice E. Shapley, Feige Wang, Qiaoya Wu, Qian Yang

First listed 2025-05-26 | Last updated 2025-05-26

Abstract

We present a spectroscopic sample of 23 broad-line AGNs (BLAGNs) at $3\lesssim z\lesssim 6$ selected using F322W2+F444W NIRCam/WFSS grism spectroscopy of the central 100 ${\rm arcmin^2}$ area of the NEXUS survey. Among these BLAGNs, 15 are classified as Little Red Dots (LRDs) based on their rest-frame UV-optical spectral slopes and compact morphology. The number density of LRDs is $\sim 10^{-5}\,{\rm cMpc^{-3}}$, with a hint of declining towards the lower end of the probed redshift range. These BLAGNs and LRDs span broad H$α$ luminosities of $\sim 10^{42.2}-10^{43.7}\,{\rm erg\,s^{-1}}$, black hole masses of $\sim 10^{6.3}-10^{8.4}\,M_\odot$, and Eddington ratios of $\sim 0.1-1$ (median value 0.4), though the black hole mass and Eddington ratio estimates carry large systematic uncertainties. Half of the LRDs show strong Balmer absorption, suggesting high-density gas surrounding the line-emitting region. We detect extended (hundreds of parsec) rest-frame UV-optical emission from the host galaxy in the majority of these LRDs, which contributes significantly or even dominantly to their total UV emission. This host emission largely accounts for the peculiar UV upturn of the LRD spectral energy distribution. We also measure the small-scale ($\lesssim 1\,{\rm cMpc}$) clustering of these BLAGNs and LRDs by cross-correlating with a photometric galaxy sample. Extrapolating the power-law two-point correlation function model to large linear scales, we infer a linear bias of $3.30_{-2.04}^{+2.88}$ and typical halo masses of a few $\times 10^{11}\,h^{-1}M_\odot$ for BLAGNs at the sample median redshift of $z\sim 4.5$. However, the inferred linear bias and halo masses of LRDs, while formally consistent with those for BLAGNs at $\sim 1.5σ$, appear too large to be compatible with their space density, suggesting LRDs may have strong excess clustering on small scales.

Short digest

Using NIRCam/WFSS F322W2+F444W in the central 100 arcmin² of NEXUS, the team secures a spectroscopic census of 23 BLAGNs at 3≲z≲6, 15 of which satisfy Little Red Dot criteria. The sample spans L(Hα)≈10^{42.2}–10^{43.7} erg s⁻¹, M_BH≈10^{6.3}–10^{8.4} M_⊙, and λ_Edd≈0.1–1 (median 0.4); about half of the LRDs show strong Balmer absorption indicating dense gas near the emission region. Most LRDs reveal extended (hundreds of parsec) rest-UV/optical host emission that significantly or dominantly contributes to the UV, explaining the characteristic UV upturn, and the LRD number density is ≈10⁻⁵ cMpc⁻³ with a mild decline toward lower z. Small-scale (≲1 cMpc) cross-correlations give a BLAGN bias of ~3.3 and halo masses of a few×10¹¹ h⁻¹ M_⊙, while LRD clustering appears overly strong for their space density, hinting at excess small-scale clustering; the single-epoch M_BH and λ_Edd carry large systematics.

Key figures to inspect

• WFSS 1D/2D spectra and line-fit panels for the BLAGN/LRD sample: verify broad Hα widths, the incidence and depth of Balmer absorption, and how continuum removal affects line profiles.
• SED/colour–slope figure for LRDs: inspect β_UV and β_opt bending and the decomposition showing how extended host light drives the UV upturn relative to the nuclear component.
• Postage stamps and surface-brightness profiles in rest-UV/optical: confirm the detected ≳100–500 pc extended emission and quantify the host’s fractional contribution to total UV flux per source.
• Two-point cross-correlation and power-law fit: read off the ≲1 cMpc excess, derived bias (~3.3) and inferred halo masses; compare LRD vs BLAGN curves to see the small-scale enhancement for LRDs.
• Distributions of L(Hα), M_BH, and λ_Edd: check the quoted ranges and median λ_Edd≈0.4, and note any flags illustrating systematic uncertainties in single-epoch virial masses.