Little Red Dots and their Progenitors from Direct Collapse Black Holes

TL;DR

This paper uses the A-SLOTH semi-analytic model to test whether Little Red Dots (LRDs) at high redshift can be progeny of heavy direct collapse black holes (DCBHs) versus light stellar-remnant seeds. By implementing DCBH seeding criteria and accretion physics, and calibrating to high-$z$ BH mass functions, the authors compare predicted LRD demographics, host halo properties, and synthetic spectra to JWST observations. They find that heavy DCBH seeds better reproduce the LRD BH mass function, number densities, and the spectral shapes of extreme LRDs, especially under scenarios of substantial dust attenuation or dense circumnuclear gas; super-Eddington growth of light seeds tends to overproduce observable LRDs and struggle with metallicity constraints. The work highlights metallicity measurements as a decisive diagnostic and suggests that a combination of seeding channels may contribute to the LRD population, with future multi-wavelength and gravitational-wave data providing critical tests of the DCBH pathway.

Abstract

The James Webb Space Telescope (JWST) has discovered a new population of objects, the Little Red Dots (LRDs), characterized by V-shaped spectra indicative of strong breaks around the Balmer limit and compact morphology that gave them their name. A popular explanation is that they are a sub-population of active galactic nuclei/supermassive black holes (AGN/SMBHs) predominantly found in the high-redshift Universe ($z\gtrsim3$). Similarly, direct collapse black holes (DCBHs), theorized to form from collapsing massive, extremely metal-poor gas clouds, have been invoked to explain high-redshift quasars, the most massive AGN sub-population. Here, we employ the semi-analytical code A-SLOTH to produce a population of DCBHs and compare them against observed LRD demographics and properties. Specifically, we compare the DCBH-seeded SMBH population against the standard stellar-remnant seeds and find that DCBH models agree better with observed LRD population statistics and host halo properties. Furthermore, for the most extreme and earliest LRD detections, interpreted to be systems with an AGN but little stellar component, DCBHs are able to reproduce the observed spectral shape and properties under multiple scenarios - high dust attenuation or AGN surrounded by dense gas - that have been proposed to explain the unique shape of LRD spectra. Even when super-Eddington accretion, invoked previously to explain the nature of LRDs, is enforced on stellar remnant seeds, the spectral characteristics of extreme LRDs cannot be reproduced. We emphasize the importance of gas-metallicity observations as an additional dimension besides the widely used SMBH-stellar mass ratios to further constrain the progenitors of LRDs.

Figures (7)

• Figure 1: LRD population demographics: mass function at $z=4-5.5$ ( top) and (massive) BH number density ( bottom). We show models for heavy seeds, heavy seeds with a stricter LW flux criterion for formation, and (super-)Eddington light seeds (i.e., light seeds accreting at or up to 1.5 times the Eddington limit), compared against observations. We include the observed broad line LRD BHMF at $z=4-5.5$Matthee2023 and the broad line AGN (BLAGN) BHMF at $z=3.5-6$Taylor2024. We also plot in open circles the approximate total BHMF based on the TRINITY model Zhang2023, which includes obscured and dormant SMBHs. The LRD BHMF at the lowest mass bin ($\sim10^7$ M$_\odot$) is shown as an open circle as only three objects were included. For the number density, we include measurements from Kocevski2025Kokorev2024_lrdZhuang2025. The heavy seed models largely agree with the LRD observations, but the (super-)Eddington light seed model overproduces the observed BHMF. Furthermore, strengthening the criteria for heavy seed formation does not significantly affect the results. Specifically, applying a higher LW flux threshold does not affect the formation and number density of the most massive DCBH-seeded BHs, which form in the most biased and massive halos (see Fig. \ref{['fig:mhalo']}).
• Figure 2: The SMBH bolometric LF models at $z=4.5-6.5$ ( top) and $z=6.5-8.5$ ( bottom), compared with LRD observations. For the lower redshift bin, we consider models for heavy seeds, including the case of heavy seeds with a stricter LW flux criterion for formation, and (super-)Eddington light seeds. We also show the QLF at $z=5$ detected in hard X-rays before JWSTShen2020. For the higher redshift bin, we additionally show models with forced super-Eddington accretion onto heavy and light seeds. We include various observed LRD LFs Kokorev2024_lrdGreene2023Greene2025Akins2024. At $z=4.5-6.5$, the QLF observed prior to JWST is lower than the LRD observations, highlighting the unexpected abundance of LRDs. The DCBH heavy seed LF agrees well with the observed LRD LF, while (super-)Eddington light seeds overproduce the LF, similar to Fig. \ref{['fig:bhmf']}. However, at $z=6.5-8.5$, while heavy seeds largely agree with the LF shape, both BH seed models cannot reach the highest luminosities. Those luminosities (near $10^{47}$ erg s$^{-1}$) can only be reached by forcing super-Eddington accretion onto heavy or light seeds, which in turn overproduces the LRD LF. This may indicate that a combination of a very small fraction of efficiently accreting heavy seeds are needed to produce the brightest objects, while most of the heavy seed population grows at a slower rate Jeon2025. We note that recent work in Greene2025 corrected their LRD luminosity measurements to lower values by around a factor of 10 in the lower redshift bin and 100 in the higher redshift bin compared to Greene2023, so that super-Eddington accretion is not necessary to reach the highest luminosity bin. All models lie above the observed lowest luminosity bin near $10^{44}$ erg s$^{-1}$. This may be due to observational incompleteness for fainter objects, or loss of lower-mass sources due to dynamical effects that are missed in the SAM (see main text).
• Figure 3: Global gas supply of LRD hosts. The host halo total ( left) and cold ( right) gas mass, for heavy DCBH and light (super-)Eddington seed models. We plot the average, median, and maximum masses for each population of BH seeds. For both the total and cold gas reservoirs, the light (super-)Eddington seed case exhibits slightly higher masses for all categories. Thus, heavy DCBH seeds are more efficient in accreting available cold gas. However, the maximum gas mass for the two kinds of SMBH seeds shows no significant difference, indicating that, rather than the overall gas supply, its distribution near the central SMBH is more important in governing BH growth trajectories, and how extreme objects like MoM-BH* could emerge.
• Figure 4: Distribution of the central halo gas density $\rho_0$ (assuming an isothermal halo profile) and the total gas mass for heavy seeds, (super-)Eddington accreting light seeds, and forced super-Eddington growth heavy and light seeds. The overall distributions for the two seeds are similar, except that the heavy (DCBH) seed models include systems with a lower central density ($\lesssim10^{-3}$ cm$^{-3}$) at low gas masses ($\lesssim10$ M$_\odot$) and higher central density ($\sim10^{4}$ cm$^{-3}$) at intermediate gas masses ($\sim10^6$ M$_\odot$). We note that the high central density systems ($>10^{3}$ cm$^{-3}$) occur at high redshifts ($z\gtrsim13$). Thus, the presence of heavy seeds can induce high gas densities at early times, but also low densities due to their efficient accretion. We also show in the blue lines the evolutionary tracks of individual systems that reached the highest $\rho_0$ values. The redshifts at the beginning and end of the tracks are indicated. For the more extreme cases of heavy seeds and forced super-Eddington growth, there are rapid phases of gas mass and $\rho_0$ changes throughout the halo's evolution due to gas accretion and outflow. At later times, when the halo is more massive and outflows are not as strong, the track settles down to a phase of increasing gas mass. For light seeds without forced super-Eddington accretion, such rapid changes are not seen, and many objects experience mergers, so that they survive for a shorter period of time. The periods of rapid gas inflow, increasing the central density and gas mass, may be necessary to produce objects like MoM-BH*, which are predicted to be gas rich with low stellar mass at $z\sim8$, long after the primordial conditions of the first BH seed formation.
• Figure 5: Metallicities of SMBH hosts at high-redshift. $M_{\rm BH}/M_*$ vs. the average metallicity for host halos of heavy DCBH seeds, (super-)Eddington light seeds, as well as forced super-Eddington heavy and light seeds for $z=5-10$. For comparison, we show select cases of lensed SMBH observations within this redshift range, Abell2744-QSO1 Maiolino2025, MoM-BH* Naidu2025, and CAPERS-LRD-z9 Taylor2025. While the overall shape of the distribution looks similar for both seed models (other than the light seeds with forced super-Eddington accretion), heavy seeds in general reside in more overmassive systems. Similarly, the observations also prefer heavy seeds to produce overmassive and metal-poor systems (or the extreme case of light seeds with forced super-Eddington growth). Thus, metallicity measurements of high-$z$ AGN could be a key diagnostic of their origins. We also show individual evolutionary tracks for the same objects as in Fig. \ref{['fig:rho0gas']} ( blue lines), again indicating their starting and ending redshifts. For the heavy and forced Eddington seeds, they initially form in metal-free systems. We therefore indicate their first timesteps as dashed lines, originating at extremely low (or zero) metallicity values. The chosen tracks represent the extreme systems that reached the highest central density values in Fig. \ref{['fig:rho0gas']}.