The importance of super-Eddington black hole accretion for the emergence of massive quiescent galaxies at high redshift

TL;DR

The paper investigates why COLIBRE simulations better match the observed abundance of massive quiescent galaxies at $z\gtrsim 6$ by testing the role of super-Eddington black hole accretion. It employs cosmological hydrodynamical COLIBRE runs with varying caps on the Eddington ratio, including a fiducial model allowing $f_{\rm Edd,max}=10^2$ and two variants with $f_{\rm Edd,max}=1$ and $0.1$. The main finding is that super-Eddington accretion accelerates early black hole growth, triggering AGN feedback earlier and producing MQGs in agreement with JWST data, with about $50\%$ of BH mass growth at high redshift arising from $f_{\rm Edd}>1$ episodes, though these events are brief in time. This supports the view that BH physics and AGN feedback modelling, including non-spherical accretion, are crucial to reproducing the early quenching of massive galaxies in a $\Lambda$CDM cosmology.

Abstract

Recent JWST observations indicate that massive quiescent galaxies (stellar mass $M_{*}\gtrsim 10^{10}~\mathrm{M_\odot}$) at high redshift ($z\gtrsim 6$) are more abundant than predicted by most existing galaxy formation simulations and semi-analytic models. Notably, the new COLIBRE simulations have succeeded in reconciling this tension, though the precise reason for their improved agreement with JWST data remains unclear. We demonstrate that the improved agreement is largely due to super-Eddington growth of supermassive black holes (BHs) at high redshift. We run a series of $(100~\mathrm{cMpc})^{3}$ simulations with the COLIBRE subgrid physics, varying the maximum allowed BH accretion rate in units of the Eddington rate. We show that only the fiducial COLIBRE model, which permits super-Eddington accretion, is consistent with the JWST constraints at $z \gtrsim 6$. Moreover, we find that in COLIBRE about $50$ per cent of BH mass growth at high redshift occurs in the super-Eddington regime, even though such events are extremely rare in time. Our work highlights the important role of super-Eddington accretion in simulations of galaxy formation for reproducing the observed early emergence of quenching of massive galaxies.

Figures (4)

• Figure 1: An example of a representative massive galaxy whose star formation is quenched by AGN feedback in the simulations with different maximum allowed Eddington fractions: $f_{\rm Edd,max} = 10^2$ (yellow), $1$ (red), and $0.1$ (blue). In all three cases, the evolution of the same galaxy is shown. From left to right, the top row shows the galaxy stellar mass, its sSFR, and its BH mass. The bottom row shows the Eddington fraction, the cumulative AGN feedback energy injected by the BH, and the cumulative fraction of mass accreted by the BH in $f_{\rm Edd} > 1$ accretion events (see the main text for details). Allowing super-Eddington accretion results in enhanced BH growth at high redshift, leading to the galaxy becoming quenched by AGN feedback at earlier times.
• Figure 2: Evolution of the sSFR (top), BH mass (middle), and the cumulative fraction of BH mass accreted at $f_{\mathrm{Edd}} > 1$, all plotted against stellar mass. Results are shown for simulations with $f_{\rm Edd,max}=10^2$, $1$, and $0.1$ (colours). Columns show different redshifts (left to right): $7.5$, $7$, $6.5$, and $6$. Solid lines indicate the median relations and shaded regions the 16$^{\rm th}$ to 84$^{\rm th}$ percentiles, all computed in $0.2$-dex $M_*$ bins. In the top panels, the black horizontal lines mark the quenching threshold of $0.2/t_{\rm age}$. Individual circles denote all MQGs satisfying $\text{sSFR} < 0.2/t_{\rm age}$ and $M_* > 10^{9.5}~\mathrm{M_\odot}$. The enhanced BH growth due to super-Eddington accretion is evident across the entire population of massive galaxies, enabling their earlier quenching by AGN feedback and leading to a larger number of MQGs.
• Figure 3: Evolution of the comoving number density of MQGs in simulations with different maximum allowed Eddington fractions: $f_{\rm Edd,max}=10^2$ (yellow), $1$ (red), and $0.1$ (blue). Solid curves show predictions after adding $0.3$ dex lognormal errors to SFRs and $M_*$; dotted curves show results without errors. Black symbols indicate JWST constraints. Only the fiducial model ($f_{\rm Edd,max}=10^2$), which allows super-Eddington BH accretion, is consistent with the JWST data at $z \gtrsim 6$.
• Figure 4: Cumulative fraction of BH mass accreted (top panel) and time spent (bottom panel) in accretion events at different Eddington fractions (colour-coded). Results are shown for the model with $f_{\rm Edd,max} = 10^2$, using all BH particles in the simulation to calculate the cumulative fractions. The white curve marks $f_{\rm Edd} = 1$ for reference. Although super-Eddington accretion ($f_{\rm Edd} > 1$) contributes more than $\approx 50$ per cent of the BH mass growth at high redshift, with $f_{\rm Edd} \gtrsim 10$ frequently reached, BHs spend only a few per cent of their time in the super-Eddington regime.