Size matters: are we witnessing super-Eddington accretion in high-redshift black holes from JWST?

TL;DR

This study investigates whether JWST-detected high-$z$ MBHs are biased toward overmassive estimates by incorporating physically motivated slim-disc accretion spectra and a BLR response that depends on the Eddington ratio. Using an agnslim-based framework and an MCMC fit to observed broad-line widths and luminosities, the authors show that MBH masses can be overestimated by up to an order of magnitude when BLR evolution is ignored, and that many sources may actually host lower-mass, higher-rate accretors with $f_{ m Edd}$ near or above unity. The fiducial slim-disc model brings MBH–host relations closer to local expectations, though degeneracies with the virial factor and BLR structure maintain substantial uncertainties, especially for spin. Overall, the work highlights the importance of detailed disc and BLR physics in MBH mass estimates at high redshift and suggests that rapid, possibly super-Eddington growth phases could be more common than previously inferred, pending further observational constraints.

Abstract

Observations by the James Webb Space Telescope of the Universe at $z\gtrsim 4$ have shown that massive black holes (MBHs) appear extremely overmassive compared to the local correlation for active galactic nuclei. In some cases, these objects might even reach half the stellar mass inferred for the galaxy. Understanding how such objects formed and grew to this masses has then become a big challenge for theoretical models, with different ideas ranging from heavy seed to super-Eddington accretion phases. Here, we take a different approach, and try to infer how accurate these MBH mass estimates are and whether we really need to revise our physical models. By considering how the emerging spectrum (both the continuum and the broad lines) of an accreting MBH changes close to and above the Eddington limit, we infer a much larger uncertainty in the MBH mass estimates relative to that of local counterparts, up to an order of magnitude, and a potential preference for lower masses and higher accretion rates, which i) move them closer to the local correlations, and ii) might indicate that we are witnessing for the first time a widespread phase of very rapid accretion.

Figures (5)

• Figure 1: $M_{\rm BH}$ estimates from the MCMC for the validation run against the MBH mass reported in the observational studies considered in this work. The black line corresponds to the 1:1 relation, with the grey shaded area 0.3 dex wide. The dots correspond to the MBHs in yue24, greene05, harikane23, maiolino23, and ubler23. In the inset we show the results obtained for the rv15 data as cyan crosses.
• Figure 2: Corner plot resulting from the MCMC validation run on J1030+0524 yue24 for the three physical parameters of the model $M_{\rm BH}$, $L/L_{\rm Edd}$ (obtained by rescaling $L_{\rm thin}/L_{\rm Edd}$ as described in Section \ref{['sec:methods']}), and $a_{\rm BH}$. The blue lines correspond to the values reported in the original work.
• Figure 3: Left panel: same as Fig. \ref{['fig:MBH']}, but for our full model, with the estimates of the entire sample shown as red dots (fiducial) and purple squares (MR18). The black dashed line is to guide the eye and corresponds to a 0.5 dex offset relative to the 1:1 relation. Right panel: Eddington ratio distribution for our fiducial model (red dots), the MR18 case (purple squares), and the validation run (black crosses) as a function of the estimated MBH mass. The thick grey dashed line corresponds to the Eddington limit.
• Figure 4: MBH mass--stellar mass relation for the source in our sample. We show the local AGN from rv15 as blue stars and orange diamonds, with the underlying shaded area correspond to the 1-$\sigma$ and 2-$\sigma$ uncertainties around the best fits to the local samples (grey and cyan for inactive and active galaxies respectively). The original data from the literature is shown as green circles, whereas our new estimates are reported as red dots (left panel) and purple crosses (right panel) for the two virial factors considered. For completeness, we also show as magenta dotted lines constant mass ratios of 0.01 and 0.1.
• Figure 5: Reconstructed spectra for 4 selected sources in our sample: CEERS_02782, JADES_000954, J0100+2802, and UNCOVER_13821. The observed spectra (obtained from the public data release of the different programs, but for the UNCOVER source, which has been extracted from the published paper) are shown as black solid lines (with the right panel showing a zoom on the H$\alpha$ line), the blue dashed, orange dash-dotted, and green dotted lines refer to our validation, fiducial, and MR18 models respectively. The cyan vertical line in J0100+2802 corresponds to $\lambda=5100\mathring{A}$ redshifted to the observer frame, which we used to constrain the models. The grey line corresponds to the power-law continuum component from the fit by yue24. All but the UNCOVER source report absolute fluxes, whereas in the UNCOVER case the flux is normalised to the luminosity at 2500Å, as done in greene24. The numbers reported in the legend correspond to the parameters employed for each model $\log (M_{\rm MBH}/\rm M_\odot)$, $\log (L/L_{\rm Edd})$, and $a_{\rm BH}$, whereas the mass estimates above each panel correspond to those in the corresponding observational papers.