Little Red Dots are Direct-Collapse Black Hole-Forming Galaxies

Daniel J. Whalen, Muhammad A. Latif, Konstantinos Topalakis, Fergus Cullen, Devesh Nandal, Sadegh Khochfar

First listed 2026-09-14 | Last updated 2026-09-10

Abstract

"Little Red Dots" (LRDs) at $4 < z < 8$ are one of the most challenging discoveries by JWST to date because their distinctive V-shaped spectra and compact morphologies (100 - 200 pc) defy conventional astrophysical interpretation. Previous attempts to explain LRDs as compact stellar systems, heavily-cocooned black holes with differential flows, supermassive stars, or more exotic objects like 'black-hole stars' either cannot show how they formed, explain the origin of the dense shells needed for the absorption features in their spectra, or account for their observed abundances or inferred lifetimes. Here we show that LRDs are simply direct-collapse black hole galaxies in which the BH is still shrouded by the massive disk that created it. Our cosmological simulations yield spectra that are good matches to those of LRDs because high densities at the center of the disk trap X-rays from the BH and produce the observed Balmer absorption features while allowing UV, optical and reprocessed IR flux to partly escape. The host galaxy forms a dense 10$^8$ M$_{\odot}$ cluster of stars with a radius of 150 pc next to the BH, consistent with observations of LRDs. Our models reproduce a wide variety of LRD spectra from typical objects like RUBIES-EGS-42046 at $z = 5.28$ to those with the strongest Balmer breaks such as MoM-BH$^*$-1 at $z = 7.76$ and those at the highest redshifts like CAPERS-LRD-z9 at $z = 9.29$.

Short digest

Whalen et al. argue that little red dots are direct-collapse black hole galaxies observed while the newly formed black hole remains embedded in the dense accretion disk that created it. Their radiation-hydrodynamic disk calculations and composite Cloudy+BPASS spectra reproduce the V-shaped SEDs, Balmer absorption, weak escaping X-rays, and compact stellar components of LRDs, including RUBIES-EGS-42046, MoM-BH*-1, and CAPERS-LRD-z9. The proposed mechanism links the characteristic LRD continuum directly to X-ray trapping in gas reaching roughly 10^10 cm^-3 near the BH, while a predominantly Pop II cluster concentrated within about 150 pc supplies the observed compact host light.

Key figures to inspect

• Figure 1. This establishes the proposed physical engine: an ultradense DCBH accretion disk with a trapped central H II region. It is the clearest visual basis for the paper's claim that X-rays are absorbed and reprocessed rather than breaking out.
• Figure 2. This shows the second essential component of the model, a predominantly Pop II stellar cluster concentrated around the DCBH within roughly 150 pc. It connects the simulated host to the compact stellar populations inferred for LRDs.
• Figure 3. This is the core observational comparison, showing spectral fits across several LRDs spanning typical objects and an extreme Balmer-break source. It demonstrates that the same embedded-DCBH framework can accommodate diversity in the observed V-shaped SEDs.
• Figure 4. The fit to CAPERS-LRD-z9 extends the model comparison to z=9.29, making it the strongest single illustration of the proposed channel at the highest-redshift end of the current LRD sample.