A radiation-hydrodynamic simulation grew a million-solar-mass black hole seed in one protocluster halo

The far-ultraviolet flux from a neighbouring galaxy suppresses early star formation in the halo, so gas accumulates instead of turning into stars ^[10]. Collapse comes at redshift 14 ^[11]. After that the depth of the halo matters: at a virial temperature near 4 x 10^4 K, gas that has been ionized stays gravitationally bound, and accretion onto the central protostars holds high enough for them to reach 5 to 9 x 10^5 solar masses ^[13][12]. Standard direct-collapse models predict about 10^5, so these protostars are five to nine times heavier ^[12][21].

Then a dense, optically thick envelope forms around the newborn black hole and feeds its accretion disk ^[15]. The route to super-Eddington accretion is radiation trapping, with photons advected inward faster than they can diffuse outward ^[16]. That same optically thick disk also produces the observable signature, since electron scattering in it broadens Halpha and the simulated sources show red continua as well ^[4][23].

Growth from roughly 10^6 solar masses at formation to about 3 x 10^7 by redshift 8 is a factor of 30 ^[22]. Black holes above 10^7 solar masses are already present in the run by redshift 10 ^[17].

The calculation followed the collapse of one halo, with the moving-mesh code AREPO ^[7]. It sits about 10 kpc from the luminous neighbour that supplies the required flux ^[9]. Several seed-forming sites appear within the same overdense environment, and the authors take that as an indication that heavy-seed formation happens naturally in protocluster regions ^[14].

The abstract reaches past one halo. The authors write that their results offer "a cosmological explanation for their abundance and properties" ^[19]. Abundance is a rate times a duration: how often such halos form per unit volume, and what fraction of a young black hole's life is spent in the enshrouded phase that makes it look like a little red dot. The paper describes that phase as short-lived and brief ^[6][24].

Matching broad Halpha and a red continuum makes the identification consistent; it does not establish it. The observational case for fast, obscured growth in these sources was already in their spectra ^[20]. What the run adds is that direct-collapse and Population III pathways had relied on idealized conditions and, according to the authors, had not been followed self-consistently in cosmological simulations ^[18]. Here the seed forms, feeds and reddens inside the same calculation that grows the protocluster around it ^[2].