First results of AMBRA: Abundant Seeds and Early Mergers as a Pathway to the First Massive Black Holes

Abstract

AMBRA combines the large cosmological volume and statistical power of ASTRID with the physically motivated gas-based black hole seeding models from BRAHMA. Motivated by JWST's discoveries of massive black holes (BHs) at $z\gtrsim 9$, AMBRA adopts a lenient heavy-seed prescription from the BRAHMA suite, allowing for the formation of $4\times 10^{4-5}\ M_{\odot}$ seeds in halos with star-forming, metal-poor gas. The seeding model is motivated by scenarios in which heavy seeds form through stellar collisions in star clusters or from the rapid growth of Population III remnants. The improved seeding model enables AMBRA to form BH seeds much earlier and more efficiently compared to ASTRID. This significantly enhances early BH growth, producing a $z=8$ BH number density more than an order of magnitude higher than that in ASTRID over the mass range $10^{5-7}\ M_{\odot}$. BHs reaching masses consistent with GN-z11 and CEERS-1019 typically originate in highly compact density peaks and undergo multiple early mergers. In these systems, $\sim50\%$ of BH masses by $z=11$ is from BH mergers, after which gas accretion becomes the dominant growth channel. Without this early merger-driven assembly, ASTRID cannot reproduce the high-mass BH detected by JWST. Our results indicate that abundant early seed formation combined with frequent mergers can explain several JWST massive BH candidates without requiring sustained super-Eddington accretion. As a testable prediction, AMBRA yields $\approx4$ LISA detectable BH merger events per year at $z\geq8$, which is three orders of magnitude higher than that in ASTRID.

Figures (13)

• Figure 1: Top row: large-scale environment of the most massive BH at $z=8$ in AMBRA (right) and the same region in ASTRID (left). The visualization shows the gas density field in a box of $8\ {\rm cMpc/h}$ per side colored by temperature, from red to blue, indicating warm to cold. The same colorbar is applied to both panels. The yellow crosses mark all the BHs with $M_{\rm BH}\geq10^{5}$$M_{\odot}$, and the red circles mark the remnants of mergers that occur during $8<z\leq9$ (there is no such merger in the plotted ASTRID region). Middle row: from left to right, we show the BH seed formation history, the BH mass function, and the BH luminosity function at $z=8$. In all three panels, red curves show AMBRA and blue curves show ASTRID. In the left panel, arrows mark the redshift of the first seed formation: $z=26.4$ for AMBRA and $z=17.3$ for ASTRID. In the middle panel, the gray band represents the adopted seed mass range ($3\times 10^{4}\leq M_{\rm seed}\leq 3\times 10^{5}\ h^{-1}$$M_{\odot}$). In the right panel, the luminosity function is compared with the observational constraints from Greene2026. Bottom row: the ratio of the quantities predicted by the two simulation (AMBRA/ASTRID) for the corresponding panels above. The dashed horizontal line marks a ratio of 1. Overall, these comparisons demonstrate that AMBRA seeds BHs more efficiently, and produces a larger population of massive BHs at high redshift than ASTRID.
• Figure 1: Resolution test for the BH seeding history in two $12.5\ h^{-1}$ Mpc boxes. We plot the comoving BH seed number density per unit redshift, $n_{\rm seed}/dz$, as a function of redshift. The green curve corresponds to the run with the same resolution as AMBRA, while the yellow curve shows a run with $\approx 6$ times higher mass resolution. Error bars represent the Poisson uncertainties. The AMBER-resolution run produces more BH seeds at high redshift by up to a factor of $\approx 2$, while the two runs converge toward lower redshift. This indicates that the overall redshift dependence of the seeding history is in general robust to resolution.
• Figure 1: The relation between $M_{\rm BH}/M_{\rm gal}$ and the properties of the local environment. For each panel, we show the results for all central galaxies with mass above $10^{9}$$M_{\odot}$ at $z=8.5$. The sample includes 247 galaxies. The red curves plot the median value. We label the Spearman correlation in the lower right corner. The panels are ordered based on the absolute values of their correlation. For an explanation of these properties, please refer to Section \ref{['sec:overmassive']} and Fig. \ref{['fig:correlation']}.
• Figure 2: The evolution of the most massive BH at $z=8$ in AMBRA (blue) and ASTRID (red). Gray points with error bars show observed high-$z$ massive BHs population: CEERS-1019Larson2023_ceers_obs, UHZ1 Bogdan2024_uhz1_obsGoulding2023_uhz1_obs, GN-z11Maiolino2024_gnz11Tacchella2023_gnz11_obs, CAPERS-LRD-z9 Taylor2025_capers_LRD_z9_obs, and GHZ9 Kovacs2024_ghz9_obs. Among these observed BHs, two of which (GN-z11 and CEERS-1019) are reproduced by AMBRA. Top: Mass evolution of the most massive progenitors in both simulations. The vertical dash lines mark the merger events. Bottom: Bolometric luminosity history (solid) of these progenitors. The dotted curves show the corresponding Eddington limits. Compared to ASTRID, the most massive BH in AMBRA grows much more rapidly at $z\gtrsim 10$, and reaches a mass of $\sim10^{7}$$M_{\odot}$ at $z=8$, which is about an order of magnitude higher than the most massive BH in ASTRID.
• Figure 3: The counterparts of GN-z11 and CEERS-1019 (the dots) compared to the entire galaxy population in AMBRA (the green pixels). The left column is plotted based on AMBRA$z=10$ data, where we search for the counterpart for GN-z11; and the right column is based on AMBRA$z=8.5$ data, where we search for the counterpart for CEERS-1019. Upper: the UV magnitude of the galaxy $M_{\rm UV, gal}$ versus the galaxy mass $M_{\rm gal}$. Lower: the galaxy SFR versus $M_{\rm gal}$. In each panel, the gray area corresponds to the observational constraints for GN-z11 and CEERS-1019 from Tacchella2023_gnz11_obs and Larson2023_ceers_obs, respectively. We use the galaxy properties, including $M_{\rm UV}$, $M_{\rm gal}$, and ${\rm SFR}$, to identify the counterparts, and color these counterparts based on the $L_{\rm bol}$ of their central BH (the lower color bar). We hereafter study the BH properties hosted by these counterparts and compare with the JWST-based measurements.