Little Red Dots: Rapidly Growing Black Holes Reddened by Extended Dusty Flows

TL;DR

The paper proposes that little red dots (LRDs) are dust-reddened broad-line AGNs whose SEDs arise from AGN emission embedded in extended dusty flows with a diffuse density profile $\rho\propto r^{-\gamma}$ ($\gamma<1$). The UV–optical SED is shaped by a gray extinction causing a red optical continuum and a flat UV spectrum, while the IR SED is governed by re-emission from relatively cool dust, shifting the energy peak toward the mid-IR. By combining an Orion-like extinction law with $A_V\approx 3$ and a broad, extended dust distribution, the model naturally reproduces the characteristic v-shaped rest-frame UV/optical SEDs observed by JWST without invoking host-galaxy light or scattered AGN emission, and it remains consistent with MIRI and ALMA constraints. The framework implies that LRDs trace an early, rapidly growing BH population in dynamically evolving, less-concentrated gas around galactic nuclei at $z>4$, with implications for BH accretion efficiency and cosmic BH growth history.

Abstract

The James Webb Space Telescope (JWST) observations have revolutionized extragalactic research, particularly with the discovery of little red dots (LRD), which we propose are dust-reddened broad-line active galactic nuclei (AGNs). Their unique v-shape spectral feature observed through JWST/NIRCam challenges us to discern the relative contributions of the galaxy and AGN. We study a spectral energy distribution (SED) model for LRDs from rest-frame UV to infrared bands. We hypothesize that the incident radiation from an AGN, characterized by a typical SED, is embedded in an extended dusty medium with an extinction law similar to those seen in dense regions such as Orion Nebula or certain AGN environments. The UV-optical spectrum is described by dust-attenuated AGN emission, featuring a red optical continuum at $λ>4000$ A and a flat UV spectral shape established through a gray extinction curve at $λ<3000$ A, due to the absence of small-size grains. There is no need for additional stellar emission or AGN scattered light. In the infrared, the SED is shaped by an extended dust and gas distribution ($γ<1$; $ρ\propto r^{-γ}$) with a characteristic gas density of $\simeq 10-10^3~{\rm cm}^{-3}$, which allows relatively cool dust temperatures to dominate the radiation, thereby shifting the energy peak from near- to mid-infrared bands. This model, unlike the typical AGN hot torus models, can produce an infrared SED flattening that is consistent with LRD observations through JWST MIRI. Such a density structure can arise from the coexistence of inflows and outflows during the early assembly of galactic nuclei. This might be the reason why LRDs emerge preferentially in the high-redshift universe younger than one billion years.

Figures (7)

• Figure 1: Extinction curves as a function of wavelength normalized by the visual extinction at 5500 $\rm \AA$: the Small Magellanic Cloud Gordon2003, starburst galaxies Calzetti_2000, the interstellar dust in Milky Way with $R_\mathrm{V}=3.1$ and 5.0 Cardelli_1989ApJ, high-redshift galaxies at $6<z<12$Markov_2024, the Orion Nebula Baldwin1991, and composite AGN spectra Gaskell_2004. Extinction curves in the absence of small-size dust grains, as observed in the Orion Nebula, are significantly flattened at $\lambda <2500~{\rm \AA}$, unlike the other models commonly used in galaxy SED fitting methodology Labbe_2023.
• Figure 2: The flux densities of dust-reddened AGNs normalized at $4000~{\rm \AA}$ (solid curves), where the incident quasar SED is adopted from a SED template model with a removal of host galaxy and dust emission contributions Temple_2021. The six different extinction laws shown in Figure \ref{['fig:ReddeningLaw']} are applied by setting a visual extinction of $A_{V}=3$ mag. The Markov's extinction curve is not used because of the lack of full wavelength coverage. The SED shaped with the Orion Nebula extinction law reproduces the characteristic v-shape SED of LRDs. The averaged photometric data of LRDs at $z\sim 5.5$ and $7.5$ are overlaid for comparison Barro_2023.
• Figure 3: The observed SEDs of LRDs across various redshifts of $4\leq z\leq 7$, where the intrinsic AGN luminosity is set to $L_{\rm bol} = 10^{46}~{\rm erg~s}^{-1}$. We adopt the composite SED template from Temple_2021, and assume the Orion-like extinction law with $A_{\rm V}=3$ mag. The 5$\sigma$ point-source imaging depths of JWST survey programs are overlaid: COSMOS-Web Casey_2023, CEERS Finkelstein_2023, and JADES-Medium/Deep Eisenstein_2023.
• Figure 4: The flux densities of LRDs normalized at $4000~{\rm \AA}$ for different values of $n_0$ and $\gamma$, where the incident SED is the same as in Figure \ref{['fig:TempleDifferentAttenuation']}. The stacked photometric data of LRDs are taken from the JADES survey Perez-Gonzalez_2024 and the COSMOS-Web survey Akins_2024, and the upper bounds of the mid- to far-IR flux densities due to non-detection in SCUBA and ALMA bands are taken from Akins_2024. Left panel: dependence of the SED on the density $n_0$. The highest density case of $n_0=10^3~{\rm cm}^{-3}$ results in overly bright rest-frame NIR, violating the MIRI F770W constraints. Lower density cases reproduce SEDs consistent with the observed data points, while the lowest density of $n_0=1~{\rm cm}^{-3}$ yields FIR SED peak inconsistent with the upper bounds from the ALMA observations. Right panel: Variation of SEDs with different density slopes $\gamma$. As the density is less concentrated ($\gamma<1$), the energy peak shifts from the NIR to the MIR regime. The feedback-free model for each case is presented with the dotted curve.
• Figure 5: Opt-NIR slope $\beta_\mathrm{opt/NIR}$ as the function of density power-law index $\gamma$ and gas density $n_0$. High density cases lead to bright rest-frame NIR emission and thus a steeper slope. The solid curve shows the boundary of $\beta_\mathrm{opt/NIR}=-1$, below which the NIR SED in our model is consistent wit the requirement from JWST/MIRI observations for LRDs. The dashed curves present constant $30\leq T_\mathrm{out}/{\rm K} \leq 120$ and the temperature of Cosmic Microwave Background (CMB) at $z=6$. Below $T_{\rm out}\simeq 30~{\rm K}$, the SED becomes inconsistent with the non-detection of LRDs with ALMA observations. The allowed parameter space is approximated as $10^{2.9\gamma+0.83} \lesssim n_0/{\rm cm}^{-3} \lesssim 10^3$.