Little Red Dots as self-gravitating discs accreting on supermassive stars: Spectral appearance and formation pathway of the progenitors to direct collapse black holes

Abstract

We propose an alternative physical interpretation and formation pathway for the recently discovered "little red dots" (LRDs). We model LRDs as super-massive stars (SMSs) surrounded by massive self-gravitating accretion discs (SMDs) that form as a consequence of gas-rich major galaxy mergers. The model provides an excellent match for numerous spectral features of LRDs, where the V-shape arises from the superposition of two black bodies, and Balmer line broadening is sourced by the intrinsic rotation of the SMD. No additional AGN, stellar, dust, or broadening component is strictly required. This results in a model with physically motivated parameters that are robust to variations in observed LRD properties. We perform MCMC fits for two representative LRD spectra, for which the full parameter posterior distributions are determined. Allowing for a compressed SMS mass-radius relation, the recovered parameters are compatible with sub-Eddington accretion in self-gravitating discs, and the recovered SMS masses of few $ 10^6$ M$_{\odot}$ imply the subsequent formation of massive black holes (BH) that squarely follow the expected BH mass--galaxy mass relation, while also predicting a cut-off luminosity of order few $10^{44}$ erg/s in quantitative agreement with current observations. While matching the abundance of LRDs is challenging, the association to galaxy mergers produces a redshift distribution that reflects observations.

Figures (6)

• Figure 1: Simple cartoon of the assembly of a supermassive disc (SMD) and supermassive star (SMS) system that appear as little red dots (LRD). Major mergers of galaxies with at least $\sim 10^8$ M$_{\odot}$ of gas trigger strong inflows that result in a compact self-gravitating disc at sub-pc scale, with typical temperatures of $\sim 4000$ K. The disc feeds an accreting SMS which radiates as a hot black body ($\sim 20000$ K).
• Figure 2: Basic spectra of the composite SMS/SMD model for different system inclinations (top panel), component temperatures (middle panel) and component sizes (lower panel). The chosen reference parameters are representative for the best fits of typical LRD spectra (see section \ref{['S:spectra-fitting']}). We highlight the different behaviours of the model by changing a single parameter at a time by the amount shown in the labels, with respect to a baseline choice represented by the black lines. In the bottom two panels, solid-lines indicate the hot SMS component and dashed-lines the cool SMD. Note that the inclination, mass and temperature of the disc become fully determined by spectral fitting once additional physical constraints are adopted (see section \ref{['S:spectra-fitting']}).
• Figure 3: Plot of the best-fit model for J0647-1045 (blue) and COS-756434 (red). The top section shows a part of the recovered posteriors from the MCMC parameter fitting, highlighting a correlation between the SMS/SMD size and temperature. The bottom two panels show the resulting model spectra plotted over the observed data, including and excluding emission peaks. Note that the height of the emission peaks is not part of the model, and is fitted automatically. The constraints in the X-ray and IR are adapted from published work Sachhi:LRD-Xrays:2025Wang:LRDdust:2025.
• Figure 4: Comparison between the differential merger rates of galaxies with a certain stellar (blue) or gas (red) mass (see text) for a mass ratio of 1. The curves are generally normalised to correspond with the high redshift tail for the observed LRD distribution 2025kocevski, and different gas (stellar) mass rates are normalized to share the same maximum. Note how the shape of the red curves match the entire observed distribution, strongly suggesting a connection between LRDs and gas rich galaxy mergers with masses $\gtrsim 10^8$ M$_{\odot}$.
• Figure 5: Full posterior for J0647-1045. The units for the parameters [$R_{\rm D},T_{\rm D},n_{\rm D},h_{\rm D}$,$\iota$,$R_{\rm S}$,$T_{\rm S}$,$f_{\rm c}$] are [pc,K,m$^{-3}$,$-$,rad,R$_{\odot}$,K,$-$].