Rapid emergence of overmassive black holes in the early Universe

Abstract

The origin of supermassive black holes (SMBHs) remains a long-standing problem in astrophysics. Recent JWST observations reveal an unexpectedly abundant population of overmassive black holes at z>4-6, where the BH masses lie far above local scaling relations and not reproduced by current cosmological models. How such overmassive black holes form and rapidly grow within young galaxies has remained unclear. Here we present fully cosmological radiation-hydrodynamic simulations that, for the first time, self-consistently follow the birth, early growth, and emergent observable signatures of SMBHs in proto-cluster environments. We find that heavy seeds of order $10^6 M_\text{sun}$ naturally form, exceeding typical theoretical expectations by an order of magnitude. These seeds rapidly develop dense, optically thick disks whose strong electron scattering produces broad H$α$ emission comparable to that seen in little red dots (LRDs). Sustained super-Eddington accretion then drives fast growth to $\sim 3 \times 10^7 ~M_\text{sun}$ by $z \sim 8$. These results provide a unified physical scenario in which LRDs correspond to a short-lived, enshrouded phase of heavy-seed formation, naturally evolving into the overmassive quasars detected by JWST and ultimately the progenitors of today's SMBHs.

Figures (11)

• Figure 1: Formation and evolution of massive seed BHs in the cosmological radiation-hydrodynamic simulation. (a–c) Gas density distributions around the target halos at representative epochs. Black asterisks indicate heavy-seed BHs formed via the collapse of supermassive stars, while red asterisks denote light-seed BHs originating from Population III remnants. The latter almost overlaps with the position of MBH1. Panels (a) and (b) correspond to the epochs when the first (MBH1) and second (MBH2) heavy seeds form, respectively. In panel (c), both MBH1 and MBH2 have migrated into a nearby massive halo and subsequently merged. Dashed circles mark the neighboring source halos that provide the strong FUV radiation required for heavy-seed formation. The radii of the circles indicate their virial radii at each snapshot. (d) Large-scale gas density distribution across the entire simulated volume (16 $h^{-1}$ Mpc on a side), centered on the target halo, shown at the final simulation snapshot. (e) Redshift evolution of the BH masses and the stellar mass of the source galaxy. The gray line shows the growth of a light-seed BH formed at $z\simeq22$, whereas the purple and green lines correspond to the two heavy seeds (MBH1 and MBH2). Circles mark the moments when the supermassive stars collapse into BHs. The blue line denotes the total stellar mass of the host galaxy. Cyan points with error bars show observed high-$z$ SMBHs Larson+2023Maiolino+2024Nat. (f) Eddington ratios of the two heavy seeds as a function of redshift. Both MBH1 (purple) and MBH2 (green) undergo short phases of super-Eddington accretion.
• Figure 1: Redshift evolution of the LW radiation intensity ($J_{21}$) at the location of the target halo. The LW flux gradually increases over time as the stellar mass in the nearby source galaxy grows and the target halo moves closer to the radiation source. The orange shaded region denotes the critical intensity range required for the formation of massive BH seeds ($J_{21,\mathrm{crit}} \simeq 10$–$2000$) Shang+2010Sugimura+2014Latif+2014. The redshifts at which the LBH, MBH1, and MBH2 seeds form are also shown.
• Figure 2: Time evolution of the most massive BH and the stellar mass of the nearby source galaxy. The solid line shows the simulated evolution of the BH mass ($M_{\mathrm{BH}}$) as a function of the host-galaxy stellar mass ($M_\ast$). Observational estimates for LRDs from recent JWST studies Harikane+2023Ubler+2023Larson+2023Maiolino+2024Kocevski+2025 and for quasars Izumi+2021 are overplotted for comparison (note that Izumi+2021 uses dynamical masses rather than stellar masses). The dashed line indicates the local $M_{\mathrm{BH}}$–$M_\ast$ relation derived from nearby galaxies Reines+2015. The blue point marks the time when the BH enters the virial radius of the host halo. Our model naturally reproduces the overmassive BHs observed in LRDs, with $M_{\mathrm{BH}}/M_\ast$ ratios up to an order of magnitude above the local relation.
• Figure 2: Redshift evolution of the halo mass in which MBH1 forms. The dashed lines indicate halo masses corresponding to virial temperatures of $8000$ and $40000$ K. The white star symbol marks the redshift at which heavy-seed formation is expected based on the semi-analytic model of Ishiyama+2025, while the orange star symbol indicates the redshift at which heavy-seed formation occurs in our radiation-hydrodynamic simulation. The delayed onset of cloud collapse allows the halo to accumulate a larger gas reservoir, leading to the formation of an extremely massive seed BH.
• Figure 3: Formation and evolution of protostars that grow into supermassive stars, which subsequently collapse into MBH1. (a, b) Projected gas density distributions at $t = 1.24~\mathrm{Myr}$ (panel a) and $t = 2.0~\mathrm{Myr}$ (panel b), where time is measured from the formation of the first protostar. Asterisks mark protostars with masses exceeding $10^3~M_\odot$. The simulation is terminated at $t = 2~\mathrm{Myr}$, corresponding to the typical lifetime of massive stars. Even at the end of the simulation, the stars remain embedded within dense circumstellar and circumbinary disks. (c–e) Time evolution of the stellar mass (panel c), accretion rate (panel d), and stellar radius (panel e) for the four most massive protostars. The stellar masses grow steadily with time, reaching final masses of several $10^5~M_\odot$. Discrete jumps in the mass evolution correspond to stellar mergers, which are accompanied by sharp increases in the accretion rate. The stellar radii expand to $100$–$1000~R_\odot$ as a result of the high mass accretion rates of $0.01$–$1~M_\odot\,\mathrm{yr^{-1}}$Hosokawa+2013.