The growth of the gargantuan black holes powering high-redshift quasars and their impact on the formation of early galaxies and protoclusters

TL;DR

This work tackles how ultramassive black holes powering high-redshift quasars impact their hosts and surrounding halos by running zoom-in simulations on the largest Millennium halo. By modifying the FABLE model to enable earlier BH seeding, mildly super-Eddington accretion, and reduced early feedback coupling, the authors form BHs >$10^{10}$ M$_\odot$ by $z\sim6$, matching the brightest observed quasars and reproducing plausible host properties like dust mass and SFR. They show that rapid BH growth drives strong, ejective feedback that expels metal-rich gas into the CGM, broadens stellar distributions, lowers central densities, and enhances Ly$\alpha$ emission while damping central X-ray/SZ signals, with multiwavelength observables offering a path to constrain quasar feedback in the early Universe. The results suggest early, powerful AGN feedback can significantly alter early galaxy and protocluster evolution and provide observable signatures for future facilities such as Lynx, ATHENA, AXIS, JWST, and ALMA.

Abstract

High-redshift quasars ($z\gtrsim6$), powered by black holes (BHs) with large inferred masses, imply rapid BH growth in the early Universe. The most extreme examples have inferred masses of $\sim \! 10^9\,$M$_\odot$ at $z = 7.5$ and $\sim \! 10^{10}\,$M$_\odot$ at $z = 6.3$. Such dramatic growth via gas accretion likely leads to significant energy input into the quasar host galaxy and its surroundings, however few theoretical predictions of the impact of such objects currently exist. We present zoom-in simulations of a massive high-redshift protocluster, with our fiducial FABLE model incapable of reproducing the brightest quasars. With modifications to this model to promote early BH growth, such as earlier seeding and mildly super-Eddington accretion, such `gargantuan' BHs can be formed. With this new model, simulated host dust masses and star formation rates are in good agreement with existing JWST and ALMA data from ultraluminous quasars. We find the quasar is often obscured as it grows, and that strong, ejective feedback is required to have a high probability of detecting the quasar in the rest-frame UV. Fast and energetic quasar-driven winds expel metal-enriched gas, leading to significant metal pollution of the circumgalactic medium (CGM) out to twice the virial radius. As central gas densities and pressures are reduced, we find weaker signals from the CGM in mock X-ray and Sunyaev-Zeldovich maps, whose detection - with proposed instruments such as Lynx, and even potentially presently with ALMA - can constrain quasar feedback.

Figures (11)

• Figure 1: Growth of black hole mass as a function of redshift for the largest black hole in each run considered in this paper. Thinner lines show the mass growth of black holes that merge into the primary progenitor in each simulation. The blue shaded area shows the region at which a black hole would exceed the median $z=0$$M_\mathrm{BH} / M_*$ relation, and the red area shows the equivalent for a ratio of $M_\mathrm{BH} / M_* = 0.15$ (the current record $z=0$ ratio). The calculation of these regions use the stellar mass from fable only for clarity. Individual points show where observations of black holes lie: open black circles show masses $>2\times10^{9}\,$M$_\odot$ estimated from the Mgii line widths KimIm2019Yang2020Wang2021, gold hexagons show data from JWST, all using H$\beta$ line widths Eilers2023Marshall2023Larson2023 apart from the Mgii measurement of a black hole in GN-z11 at $z=10.6$Maiolino2023a. Error bars show a representative $0.5$ dex error, from the systematic uncertainty in estimating virial black hole masses from observed line widths McLureDunlop2004VestergaardOsmer2009. The black square highlights the data from J0100+2802, the brightest observed high-redshift quasar. The changes we made in our Reference simulation allow us to grow a black hole that matches both the highest redshift observed quasars and the brightest known high-redshift quasar.
• Figure 2: Evolution of black hole accretion and observability properties for the fable (left) and Reference (right) simulations. Eddington fraction (top row) and bolometric luminosity (second row) are shown for the largest black hole and its largest progenitor. Faded lines show the value at every timestep the black hole particle is active, and the darker lines show a rolling mean with a width of 100 timesteps. Observational data points are shown as either open circles or gold hexagons (for data from JWST), with errors where available, from the sources cited in Section \ref{['Section:BHMass']}. The dashed horizontal line shows the Eddington limit, $f_\mathrm{Edd} = 1$. The black squares highlight the data from J0100+2802, the largest observed high-redshift quasar. The fable cannot grow black holes big enough to explain the luminosity of very bright, high-redshift quasars, but our Reference run can. Third row: distribution of hydrogen column densities along $\sim\!10,000$ rays of length 5 kpc from the black hole. Median values are shown as solid squares for each run, compared to the empty squares for the NoAGN run. The full distribution, illustrated with a violin plot, is only shown for alternate snapshots for clarity. Dotted horizontal line shows the threshold for the gas being Compton thick ($N_\mathrm{H} = 1/\sigma_\mathrm{T}$). The Reference run is more effective than fable at clearing out gas from the centre of the quasar host galaxy. Fourth row: Dust mass within 1.5 kpc, see Section \ref{['Section:Obscuration']}, for each run with black holes (solid dark green circles) and for the NoAGN run (empty light green circles), compared to observed values (open black circles) from Gilli2022 and a measurement from J0100+2802 from Tripodi2023 (open circle surrounded by black square). The Reference run has a dust mass close to the observed value of J0100+2802, in addition to having a consistent black hole mass and luminosity. Fifth row: Distribution of apparent UV magnitudes at 1450 Å, along $\sim\!10,000$ sightlines from the black hole, see Section \ref{['Section:Obscuration']}. For $z \lesssim 6.3$ the Reference run is brighter than fable due to earlier and more effective feedback. Bottom row: Probability of observing the quasar along a given sightline with three upcoming wide-field instruments. The quasar in the Reference run has a significantly higher chance of being observed, especially for $z \lesssim 6.3$, due to more effective feedback reducing obscuration compared to fable.
• Figure 3: Mollweide projection maps of the central 5 kpc of the central galaxy at $z=6.1$ in the NoAGN (left), fable (centre) and Reference (right) runs. We show the hydrogen column density (top) and UV optical depth $\tau_\mathrm{UV}$ (bottom), both of which are calculated along the $\sim\!10,000$ rays described in Section \ref{['Section:Obscuration']}. The increase in the strength of feedback can clearly be seen going from left to right, with more low-density channels in the surrounding gas.
• Figure 4: Top: Star formation rate as a function of redshift for all simulations. Note that the AGN feedback in the fiducial fable model has a negligible effect on star formation until much later than the Reference run. Observational data points are shown as open circles, from the sources cited in Section \ref{['Section:BHMass']}, with errors where available. As a reminder, most of these black holes have inferred masses in excess of $2\times10^9$ M$_\odot$, with the exception of a few sources at $z>7$. The black squares highlight data from J0100+2802, the largest observed high-redshift quasar, with the higher SFR coming from KimIm2019 and the lower SFR coming from ALMA dust measurements Tripodi2023. The Reference run has a SFR consistent with the accurate ALMA-based SFR of J0100+2802. Middle: Evolution of the stellar metallicity of the quasar host galaxy, calculated as the mass-weighted average metallicity of all stars within twice the stellar half-mass radius. The Reference run has considerably lower stellar metallicity in the central galaxy, due to both the lower stellar mass and expulsion of metal-rich gas. Bottom: Evolution of the galaxy gas metallicity, again calculated as the mass-weighted average metallicity of all gas cells within twice the stellar half-mass radius. Similarly to the middle row, the Reference run has a lower average gas metallicity than the NoAGN or fable runs from $z=7.5$ onwards.
• Figure 5: Ratios of stellar mass to (expected) baryonic mass as a function of halo mass for the three runs in this work, showing their evolution in this parameter space. The values for each run at $z=9,8,7$ and $6$ are shown by the star, triangle, square and circle symbols, respectively. We also show the median ratio of this quantity from a number of models (though we note these all assume Planck cosmology) from Behroozi2013Behroozi2019Moster2018. The Reference run is clearly more effective at regulating star formation compared to fable or a run without black holes.