Black Hole Envelopes in Little Red Dots

TL;DR

This paper tackles the paradox of Little Red Dots (LRDs) — high-z, compact, red AGN-like objects with broad Balmer lines but weak X-ray/radio signals — by showing that standard super-Eddington outflows would produce strong momentum and energy feedback that should disrupt their gas. The authors propose a dense, optically thick BH envelope that gravitationally confines the outflow and reprocesses its energy into a BH-powered photosphere, yielding a red optical continuum with a characteristic $T_{\rm ph}\sim 5000$–$7000$ K and a Hayashi-track–like envelope evolution. The envelope remains sustained by ISM inflow and can grow until the BH reaches a critical mass, connecting LRDs to the early rapid growth of supermassive black holes. The model accounts for the observed SEDs and variability timescales while suppressing outflow feedback, though it relies on simplified 1D hydrostatic physics and invites future 3D radiative-transfer studies to refine UV ionization and BLR formation aspects. Overall, the envelope scenario provides a cohesive framework linking LRDs to early SMBH assembly and feedback regulation in the young universe.

Abstract

Recent observations by the James Webb Space Telescope have uncovered a population of compact, red object ($z\sim 4\text{--}7$) known as little red dots (LRDs). The presence of broad Balmer emission lines indicates active galactic nuclei powered by supermassive black holes (BHs), while LRDs exhibit unusually weak X-ray and radio emission and low variability, suggesting super-Eddington accretion that obscures the central engine. We suggest that such an extreme accretion disc inevitably drives strong outflows, which would disrupt the LRDs themselves unless confined within the nuclear region -- posing a general feedback problem for overmassive BHs. To resolve this, we propose that the BH is embedded in a massive, optically thick envelope that gravitationally confines the outflow, making any outflow a no-go. This envelope, powered by accretion on to the BH, radiates at nearly the Eddington limit, and is sustained by an infall of the interstellar medium at a rate on the order of $\sim 1 M_{\odot}~{\rm yr}^{-1}$. A photosphere emerges either within the envelope or in the infalling medium, with a characteristic temperature of $5000$ - $7000 {\rm K}$, near the Hayashi limit. The resulting blackbody emission naturally explains the red optical continuum of the distinct V-shaped spectrum observed in most LRDs. Furthermore, the dynamical time-scale at the photosphere, $\sim 0.01~{\rm pc}$, is consistent with the observed year-scale variabilities. The nuclear structure and spectral features of LRDs are shaped by this envelope, which not only regulates feedback but also acts as a gas reservoir that sustains rapid BH growth in the early universe.

Figures (5)

• Figure 1: The BH mass $M_{\rm BH}$ versus stellar mass $M_{*}$ relation compared with the too-strong-feedback condition (gold solid line) in equation (\ref{['eq:feedback']}) using the outflow efficiency $\xi=0.03$ in equation (\ref{['eq:dotPout']}), the ratio of the outflow time to the Salpeter time $f_{\rm Sal}=t_{\rm out}/t_{\rm Sal}=0.3$, the ratio of the gas mass to the stellar mass $\zeta_{\rm gas}=M_{\rm gas}/M_{*}=0.1$, the Eddington ratio $\dot{m}=1$ and the LRD radius $r_{\rm LRD}=100~\mathrm{pc}$. The gold dashed line indicates the same criterion adopting $r_\mathrm{AGN} = 3\,\mathrm{kpc}.$ The dashed black lines indicate loci of constant BH to stellar mass ratio, $M_\mathrm{BH} / M_* = 0.001, 0.01, 0.1$ respectively. A large fraction of LRDs would be expected to be strongly affected by the outflow if such an outflow from a super-Eddington disc existed. The data are taken from Kocevski2025ApJ and based on Kocevski2023ApJReines+V:2015Izumi+:2021Harikane+:2023Maiolino2024AA. Note that these BH masses are estimated with the local scaling relations based on Balmer emission line FWHM and luminosity provided by Reines+V:2015Greene2005ApJ. The two X-ray detected LRDs are marked with an X Kocevski2025ApJ.
• Figure 2: Concept of the BH envelope. The BH envelope serves to block and gravitationally confine the outflow from the super-Eddington disc. Meanwhile, energy is transported through radiation and convection, and radiated away from the envelope surface. The luminosity is primarily determined by the mass accretion rate on to the BH, $L \sim \eta \dot{M}_{\rm BH} c^{2}$, and is assumed here to be regulated to the Eddington luminosity of the entire system. Mass inflow from the interstellar medium (ISM) into the envelope $\dot{M}_{\rm ISM}$ is also considered.
• Figure 3: Minimum envelope mass required to be gravitationally bound for an Eddington luminosity $L=L_{\rm Edd}=10^{45}~\mathrm{erg~s^{-1}}$, as a function of the envelope radius (red solid line). The maximum envelope mass is also shown, for which its Eddington luminosity $L_{\rm env}$ is less than $L$ (green dotted line). The radius corresponding to a photospheric temperature of $T_{\rm ph}=7000~{\rm K}$ is also plotted (blue dashed line). We also present the envelope solution obtained in section \ref{['sec:structure']} and the solution derived in Ulmer1998AA. We adopt $\beta_c = 0.01$ here. Because of the beginning of hydrogen recombination, the solutions approach the same track, which resembles the behaviour of a massive star, the Hayashi track. Ulmer's solution corresponding to $\beta=10^{-5}$ is located entirely below $1M_\odot$ and is gravitationally unbound.
• Figure 4: The envelope mass as a function of the effective temperature for a BH mass of $10^7M_{\odot}$. Solid and dashed curves correspond to the cases with a convection efficiency of $\beta_c=0.1$ and $0.01$, respectively. Here, the toy opacity model in Begelman2008MNRAS is used (see equation \ref{['eq:opc']}).
• Figure 5: SED of stacked LRDs with theoretical curves obtained by our model. Stacked SED data points are taken from Table 4 in Akins2024arXiv. Each line shows the envelopes with different photospheric temperatures and $A_V$. The green and grey dashed line highlights the blackbody component with $T = 7000~\mathrm{K}$ and $A_V = 0, 3$ respectively. We adopted a simple SED model that includes the star-formation effect as a power law and set the Lyman break around $0.1$ micron artificially since our work focuses on the redder part of the SED. The dust attenuations are included by the Calzetti dust extinction law Calzetti2000ApJ.