Heavy Seeds and the Multi-scale Fueling of Early Supermassive Black Holes and Little Red Dots

abstract · abstract

The formation of supermassive black holes (SMBHs) in the early Universe has been a longstanding mystery due to the brief period available to assemble them. More recently, the unexpected discovery with the James Webb Space Telescope (JWST) of Little Red Dots (LRDs), compact sources exhibiting signatures of active galactic nuclei (AGN), has accentuated this challenge. Cosmological simulations investigating their formation and evolution confront the limitation that the compact scale of SMBH/LRD physics cannot be simultaneously resolved with the larger-scale cosmological environment. We address this limitation by utilizing a cutout technique, selecting a targeted region around an SMBH seed from a cosmological parent simulation and re-running the cutout at higher (sub-pc) resolution for shorter periods. We explore two SMBH formation scenarios: the standard, light seed stellar-remnant case, and the heavy-seed, direct collapse black hole (DCBH) channel. Within the high-resolution achieved by the cutouts, the DCBH, due to its larger mass, is able to induce a buildup of high density gas nearby and accrete gas in bursts of super-Eddington phases, growing in mass and producing an LRD-like spectrum. In contrast, the stellar remnant cannot do so and remains dormant, unable to affect the nearby gas or significantly evolve by gas accretion alone. Therefore, we find that observed high-redshift SMBHs and LRDs require the existence of heavy seeds to assemble dense gas around themselves, enabling efficient growth and producing red optical spectra with strong Balmer breaks resembling LRDs.

S2.SS1: II.1 Simulation setup / block 1 · html

The cutouts (see Sec. II.1.1 for details) are run for ≲ 10 \lesssim 10 Myr. We thus avoid unphysical edge effects, resulting from the artificial outer (vacuum) boundary conditions of the cutout region, where all external material has been removed. We can estimate the period over which this approximation is valid with the sound crossing time to traverse the cutout region. Given an average gas temperature of ∼ 10 3 − 10 4 \sim 10^{3}-10^{4} K, corresponding to a sound speed of ∼ 10 − 5 \sim 10^{-5} pc yr -1 , the distance corresponding to the 10 Myr period is ∼ 100 \sim 100 pc, half the size of the cutout region. We accordingly limit our analysis to < 8 <8 Myr periods. Table 1 summarizes the simulation setup and Figs. 1 and 2 illustrate the wide range of length scales involved. We further verify the validity of the cutout methodology with an idealized test case of a collapsing spherical gas cloud, as described in Appendix A .

S3: III Results / block 1 · html

After running the zoom-in and cutout simulations, we post-process the simulation output with the yt [ 146 ] and Meshoid [ 48 ] packages. We compare the gas environment at high resolution for BHs with different origins, see how BH accretion and feedback behaves close to the BH, and predict their spectra from the accretion and nearby gas environment.

S3.SS1: III.1 Gas distribution near the Black Hole / block 1 · html

Comparing the resulting gas temperature distribution in Fig. 5 , we find no significant differences in the temperature range between the DCBH and stellar remnant cutouts, both exhibiting average temperatures of ∼ 10 3 \sim 10^{3} K. We note that for the DCBH case, in difference from the substantially higher densities reached, the temperatures of the gas coalescing close to the BH are similar to those of the more widely distributed gas. In fact, the DCBH cutout region contains higher temperature gas ( > 10 4 >10^{4} K) from SNe in the original Zoom simulation. Still, despite the presence of such hotter gas, the DCBH is able to trigger a dense gas buildup.

S3.SS1: III.1 Gas distribution near the Black Hole / block 1 · html

To assess whether numerical resolution has any effect on the results, we examine the gas distribution in the DCBH_Highres cutout. Fig. 6 shows the gas density and temperature distributions versus the distance to the BH in the DCBH_Highres run. The gas behaves similarly to the lower resolution case, except that the gas density near the BH reaches even higher values, up to ∼ 10 8 \sim 10^{8} cm -3 . The higher resolution is able to resolve the buildup of dense gas even closer to the central BH seed, but overall, the difference in resolution does not significantly affect the resulting gas distribution.

S3.SS2: III.2 Black hole accretion and feedback / block 1 · html

Figure 8: Model spectra for the case of a strongly-accreting DCBH. We use results from the DCBH_Accrete cutout run, taken at ∼ 5.7 \sim 5.7 Myr, during a super-Eddington accretion episode as input for Cloudy . The spectral calculation uses the BH accretion rate and close-in gas structure from the cutout, resolved to ∼ 0.1 \sim 0.1 pc. We compare our prediction against normalized observed LRD spectra [ 39 , 37 , 142 , 110 ] , as well as a model DCBH spectrum from previous work, where the gas distribution near the BH was not properly represented, given a resolution of ∼ 300 \sim 300 pc [ 70 ] . With the cutout technique, on the other hand, the high-density region around the BH can be more fully resolved, which produces the strong Balmer breaks seen in observed LRD spectra.

S3.SS3: III.3 AGN Spectra / block 1 · html

The resulting spectra are shown in Fig. 8 . We further show in comparison the model spectrum from a zoom-in cosmological simulation with lower resolution [ 70 ] , as well as normalized observed LRD spectra, including the BH* object [ 110 ] , CAPERS-LRD-z9 [ 142 ] , and RUBIES-EGS-49140, with spectral data from the CANDELS-Area Prism Epoch of Reionization Survey [ 39 , 37 , CAPERS;] . Whereas both simulation spectra show a red AGN continuum, the strong Balmer-break feature, similar to that of observed LRDs, can only be seen when the high-density gas in the vicinity of the central BH is properly resolved. Here, the required high resolution is achieved through the cutout methodology.

S4: IV Discussion / block 1 · html

On the basis of the BH evolution resulting from the cutout simulations, we consider the implications for the overall SMBH mass evolution and the possible origin of the compact LRDs.

S4.SS1: IV.1 Supermassive Black Hole Formation / block 1 · html

This conclusion would imply that most of the observed high-redshift SMBHs originated from heavy seeds. While other channels such as BH mergers exist for BH mass growth besides gas accretion, previous studies have found that accretion significantly contributes to SMBH growth even when considering high merger rates [ 16 , 17 ] . Furthermore, there are multiple pathways for heavy seed formation in addition to the DCBH scenario, which is our main focus here, such as hierarchical mergers (see Sections 1 and IV.3 ). If the various heavy seed pathways place the BH into a largely similar environment, allowing for a period of time to lose memory of the specific seeding process, our cutout simulations would then be expected to exhibit similar behavior for the subsequent evolution, provided that the main factor determining the gas distribution and the resulting BH accretion is the BH mass, not the seed formation mechanism. However, the DCBH pathway naturally places a massive seed in a pristine and cold gas environment lacking stars, and may be more effective in inducing dense gas buildup and super-Eddington accretion. Scenarios like hierarchical mergers, on the other hand, will form heavy seeds within dense stellar environments, such that the gas may be heated and/or expelled by stellar feedback even in the presence of a deeper BH gravitational potential well.

S4.SS3: IV.3 Comparison with Previous works / block 1 · html

We compare our results to select studies with a similar focus on early BH growth in a cosmological context, without aiming for completeness, to gauge the results’ robustness in terms of numerical methodology and different assumptions for the sub-grid physics.

article / block 1 · html

Our results demonstrate that heavy-seed formation is a natural outcome of early structure formation in overdense regions, with profound implications for the observed abundance of early quasars and their gravitational-wave signatures. The overmassive BHs produced in these simulations represent plausible progenitors of the first quasars, bridging the gap between the birth of seed BHs and the luminous active nuclei observed by JWST.

article / block 1 · html

The LW background is primarily contributed by Pop II stars, and its amplitude depends on assumptions regarding the IMF and the stellar models adopted. We note that uncertainties in the LW-intensity modeling have only a minor impact on our results, as we later test explicitly. A halo begins to form Pop III stars once its mass exceeds the critical threshold determined by the local LW intensity ( Kulkarni et al., 2021 ) . Under strong LW irradiation, star formation is delayed until the halo virial temperature reaches ∼ 8000 \sim 8000 K, at which point Ly α \alpha cooling triggers rapid collapse. Such halos typically experience high mass accretion rates and host more massive Pop III stars, extending to supermassive stars (SMSs) with M ∗ ∼ 10 5 ​ M ⊙ M_{\ast}\sim 10^{5}~M_{\odot} , which subsequently collapse into heavy black-hole (BH) seeds. Indeed, Ishiyama and Hirano (2025) identified more than ∼ 10 4 \sim 10^{4} SMSs with the masses larger than 10 5 ​ M ⊙ 10^{5}~M_{\odot} as heavy seed candidates in their simulation volume.

S0.SS0.SSS0.Px1: Introduction and Summary.— / block 1 · html

The James Webb Space Telescope (JWST) has recently discovered a class of compact, red-hued galaxies at high redshifts ( 4 ≲ z ≲ 11 4\lesssim z\lesssim 11 ), dubbed “little red dots” (LRDs) [ 1 , 2 , 3 , 4 , 5 ] , for which the origin remains open. Moreover, around 80% of these galaxies display broad Balmer emission lines, suggesting that they host active galactic nuclei (AGNs) with central supermassive black holes (SMBHs) of 10 5 ​ – ​ 8 ​ M ⊙ 10^{5\textup{--}8}{\rm\,M}_{\odot} [ 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 ] .

S0.SS0.SSS0.Px1: Introduction and Summary.— / block 1 · html

Fig. 1 summarizes the parameter space in terms of the modulation strength | η | |\eta| and the comoving wavenumber ratio k l / k s k_{l}/k_{s} , where PBH clusters can form and potentially explain the origin of LRDs (light-gray region) via runaway mergers. The remaining regions represent parameter ranges where SMBHs at high redshift cannot form through mergers, as will be discussed in the following sections.

S3.SS1: III.1 Spectral fitting / block 1 · html

This procedure results in a good fit for one of the objects, CFHQS J2229+1457, which exhibits pronounced narrow emission lines, while we need to implement further constraints to fit the rest of the spectra due to their relatively weak [O III ] lines that blend with the enhanced [Fe II ] emission. In particular, for all other quasars we model the complex line profiles of H α \alpha and H β \beta with two broad Gaussian components instead of one, and we also model the [S II ] λ ​ λ \lambda\lambda 6718 , 6732 \,6718,6732 doublet (with shared linewidths and velocity offsets but without a fixed line ratio) in order to accurately fit the red wing of the broadest H α \alpha component. We further tie the linewidths and velocity offsets of the narrow H α \alpha component to the single-Gaussian components of the [N II ] λ ​ λ \lambda\lambda 6549 , 6585 6549,6585 and [S II ] λ ​ λ \lambda\lambda 6718 , 6732 6718,6732 emission lines. Similarly, we tie the narrow component of H β \beta to the narrow components of [O III ] λ ​ λ \lambda\lambda 4959 , 5007 4959,5007 . The extracted quasar spectra as well as the resultant spectral fits are displayed in Fig. 1 .

S4.SS2: IV.2 Search for extended emission / block 1 · html

Table 4: A summary of H α \alpha and [O III ] nebular (non-)detections and their properties: the 5 ​ σ 5\sigma detection limit over a 1 ​ arcsec 2 1{\rm arcsec}^{2} aperture, SB 5 ​ σ 1 {\rm SB}^{1}_{5\sigma} , the maximum extent of the nebula, d max Neb d_{\rm max}^{\rm Neb} , and the corresponding lifetime measurement, log ⁡ t Q Neb \log{t_{\rm Q}^{\rm Neb}} .

S2.SS4: II.4 Black hole feedback / block 1 · html

We implement BH feedback through thermal energy injection, resulting from BH accretion. The energy injected to the nearby cold gas at each timestep δ ​ t \delta t is determined as

S3: III Results / block 1 · html

We compare our model outputs with various LRD observations, specifically considering two classes of BH populations: heavy DCBH seeds, and light stellar-remnant seeds that are accreting at Eddington/super-Eddington rates. For the heavy seed models, we consider 4 cases - the Eddington-limited case, super-Eddington limited growth (up to 1.5 times Eddington), forced super-Eddington growth (where all DCBHs accrete at super-Eddington rates when gas is available with f edd = 1.5 f_{\rm edd}=1.5 ), and the case imposing a higher LW flux threshold for DCBH formation (up to J crit = 3000 − 10000 J_{\rm crit}=3000-10000 ). The default heavy DCBH case used for our analysis is the Eddington-limited model, unless stated otherwise. For the light seeds, we consider 2 cases - super-Eddington limited (up to 1.5 times) and forced super-Eddington growth ( f edd = 1.5 f_{\rm edd}=1.5 ). The default model for light seeds is the super-Eddington limited case unless stated otherwise.

S3: 3 Discussion and Conclusions / block 1 · html

The spectral index of − - 0.85 is typical of optically-thin synchrotron emission and favors a non-thermal mechanism to produce the radio emission and possibly energize the LRDs. The radio source is unresolved at a scale of 0 ​ . ​ 2 ′′ 0\hbox to0.0pt{.\hss}{{}^{\prime\prime}}2 (< 600 pc at d = 700 Mpc).

S3: 3 Discussion and Conclusions / block 1 · html

Another important result is the evidence of significant time variability in the radio emission. Correcting the 5.0 GHz flux density of 2010 by a factor ( 5.0 / 6.0 ) − 0.85 (5.0/6.0)^{-0.85} to compare with the 2018 flux density at 6.0 GHz, we find that the flux density has increased by a factor of 2.3 ± \pm 0.2 over this 7.5 yr time period. A detectable variability over a timescale of years is not uncommon in AGNs and again points to a SMBH, but as we will see, another explanation is favored. This important change in radio flux density contrasts with the lack of variability at r-band reported for J1047+0739 at the 3-4% level by Burke et al. (2025).

PDF page 15 / block 2 · pdf

inally, we estimated the black hole mass (MBH) from the width and luminosity of the Hα BLR compo- nent, following the recipe of Greene & Ho (2005) and Reines et al. (2013). The resultant values (MBH = 107.8−9.1M⊙) overlap with the distributions of low- z SDSS quasars (e.g., Rakshit et al. 2020; Toba et al. 2021) and UV-luminous quasars at z ∼ 6 (e.g., Yang et al. 2021), but are at the lower end on av- erage. The Eddington ratios are in the range from λEdd = 0.5 to 1.1, suggesting that sub-Eddington to Ed- dington accretion takes place in the BL objects. Figure Figure 8. Black hole masses and bolometric lumi- nosities of luminous quasars (open squares; Willott et al. 2010; Shen et al. 2019; Yang et al. 2021, 2023) and SHEL- LQs low-luminosity quasars (grey dots; Onoue et al. 2019; Matsuoka et al. 2019b; Onoue et al. 2024, M. Onoue et al., in prep.) at z ∼ 6 −7. The dark blue sym- bols represent JWST BHEs at z > 4 (Harikane et al. 2023; ¨Ubler et al. 2023; Furtak et al. 2024; Greene et al. 2024; Juodˇzbalis et al. 2024, 2025; Maiolino et al. 2024b; Matthee et al. 2024; Schindler et al. 2024; Lin et al. 2025). The dotted lines represent constant Eddington ratios of λEdd = 1, 0.1, and 0.01, from top to bottom. 8 compares the estimated values with those of classical quasars and other JWST BHEs,4 demonstrating that the present objects bridge the two AGN populations. We also note the overall similarity of the present tar- gets with the SDSS type-II quasar candidates at lower redshifts (Alexandroffet al. 2013; Greene et al. 2014), both of which are characterized by black hole masses of MBH ∼108−9M⊙with sub-Eddington accretion, mild dust obscuration (AV ∼0 −3 mag), and high line lumi- nosities (L[OIII] ∼1043−45 erg s−1). However, we caution that our estimates reported above have large uncertainties. Black hole masses derived from single-epoch spectra are associated with statistical uncertainty of 0.3 – 0.5 dex (e.g., Vestergaard & Peterson 2006). The bolometric correc- tion to Hα luminosity has been derived from low-z AGNs, and we do not know how accurate it is when 4 This figure is a collection of heterogeneous measurements, since different works used different assumptions

PDF page 20 / block 3 · pdf

nding the whole picture of SMBH activity during cosmic dawn. 4. SUMMARY We present the results from JWST/NIRSpec observa- tions of 13 objects in the EoR. The sample was originally selected by the HSC-SSP imaging survey over >1000 deg2, and all the objects have luminous Lyα emission

S4: IV Summary / block 1 · html

In this work, we propose a novel cosmological framework linking the quasi-periodic variability of LRDs to distance measurements via a physically motivated P P – L L – T eff T_{\rm eff} relation for idealized radiation-pressure-supported envelopes. Applying this framework to the lensed LRD R2211-RX1, we obtain a preliminary constraint of H 0 = 120.7 − 46.5 + 47.0 ​ km ​ s − 1 ​ Mpc − 1 H_{0}=120.7^{+47.0}_{-46.5}\ \mathrm{km\ s^{-1}\ Mpc^{-1}} . Although the uncertainty remains large, it is dominated by the limited constraint on the pulsation period and serves as a critical proof-of-concept. Our forecast indicates that 10 years of monitoring can reduce H 0 H_{0} statistical uncertainty to 3–20%, depending on the intrinsic period.

S4: IV Summary / block 1 · html

The ultimate accuracy of this framework depends on quantifying systematic uncertainties in the P P – L L – T eff T_{\rm eff} relation, which remain to be fully characterized. Empirical calibration via low-redshift LRDs and improved modeling are essential to mitigate these effects and validate the theoretical baseline. Once calibrated, this method will provide a robust, independent H 0 H_{0} probe. Though nascent, this approach offers a complementary test for cosmological models, particularly at high redshifts where standard candles are scarce.

S4: IV Summary and Conclusion / block 1 · html

Successful Matching of Observed Spectra: By applying physically motivated broadening mechanisms, including a powerful stellar wind ( v ∞ ≈ 1500 ​ km ​ s − 1 v_{\infty}\approx 1500\,\mathrm{km\,s^{-1}} ) and large-scale macroturbulence ( σ v ≈ 1250 − 1500 ​ km ​ s − 1 \sigma_{v}\approx 1250-1500\,\mathrm{km\,s^{-1}} ), our model quantitatively reproduces the broad line profiles observed in both LRD targets, matching the measured FWHM of MoM-BH*-1 to within 4%.

S4: IV Summary and Conclusion / block 1 · html

We have presented a first-principles investigation into whether Population III supermassive stars (SMSs) can serve as the central engines for the enigmatic class of objects known as Little Red Dots (LRDs). By developing a novel pipeline that combines a GENEC stellar evolution model with a detailed radiative transfer and spectral synthesis code, we have demonstrated that a single, non-rotating 10 6 ​ M ⊙ 10^{6}M_{\odot} SMS can successfully reproduce the key observational features of two prominent LRDs: MoM-BH*-1 at z = 7.76 z=7.76 and The Cliff at z = 3.55 z=3.55 .