SHELLQs-JWST Unveils the Host Galaxies of 12 Quasars at z>6

TL;DR

This study uses JWST/NIRCam imaging in F150W and F356W to decompose quasar and host galaxy light for 12 quasars at $6.0<z<6.4$, achieving $11/12$ host detections in F356W and $7/12$ in F150W. By employing a PSF-library based two-dimensional decomposition and a 40-configuration combining strategy, the authors derive robust host stellar masses ($10^{9.5}-10^{11.0} M_\ ext{odot}$) and effective radii ($0.5-3$ kpc), finding that most hosts follow the star-forming galaxy size–mass relation at $z\sim6$, while the most massive systems are ultra-compact and resemble $z\sim4$–$5$ quiescent galaxies. Two post-starburst hosts show high stellar mass densities consistent with early quiescence, supporting a compaction-driven evolutionary pathway during the reionization epoch. The environments suggest a mix of isolated and small-pair systems, with no universal need for major mergers to fuel AGN, highlighting the role of secular processes and minor interactions. Overall, moderate-luminosity quasars provide a representative and effective probe of SMBH–host coevolution and structural transformation at the highest redshifts, offering crucial empirical constraints on galaxy compaction and quenching in the early universe.

Abstract

The advent of JWST has opened new horizons in the study of quasar host galaxies during the reionization epoch (z>6). Building upon our previous initial measurements of stellar light from two quasar host galaxies at these redshifts, we now report the detection of the stellar light from the full Cycle 1 sample of 12 distant moderate-luminosity quasar (M1450>-24 mag) host galaxies at z>6 from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP). Using JWST/NIRCam observations at 1.5 and 3.6 um combined with 2D image decomposition analysis, we successfully detect the host galaxies in 11 of the 12 targets, underscoring the high detection rates achievable with moderate-luminosity quasars. Based on two-band photometry and SED fitting, we find that our host galaxies are massive, with logM*/M_sun = 9.5-11.0. The effective radii range from 0.6 to 3.2 kpc, comparable to the sizes of inactive galaxies with similar masses at z~6 as measured with imaging from COSMOS-Web.Intriguingly, the two quasar hosts with post-starburst features, which reside at the high-mass end of our sample and exhibit relatively compact morphologies, have similar size and stellar mass surface densities to quiescent galaxies at z~4-5. These findings suggest that the so-called galaxy compaction scenario is already in place at the reionization epoch, in which gas inflows during starburst phases drive centrally concentrated star formation followed by rapid quenching, bridging the structural transition of massive galaxies from relatively extended star-forming disks to compact quiescent systems.

Figures (11)

• Figure 1: The redshift-$M_{1450}$ distribution of the SHELLQs quasars in this study (red) and other major JWST observations for luminous quasars: EIGER (blue; Yue2024) and ASPIRE (orange; Wang2023ApJ...951L...4W). The SHELLQs quasars, whose host detections were reported in our earlier papers Ding2023Onoue2024 are indicated by larger open circles.
• Figure 2: Quasar-host decomposition of NIRCam/F356W images for all twelve targets. The rows are as follows from top to bottom: (1) original science image (data), (2) data minus quasar model, i.e., the host galaxy and neighbors, (3) best-fit host Sérsic model(s), and (4) Signal-to-noise ratio (SNR) maps of the host significance; the noise is a combination of that from the data and the scatter of the corresponding top-five PSFs. In Row (4), we show the elliptical apertures within which we calculate the host SNR, as reported in Table \ref{['tab:F356W_result']}. When drawing these apertures, we use solid lines to highlight those with confirmed host detections. The inferred host-to-total flux ratios from the best fit are also indicated in the bottom panels. Our simulation results indicate that even in challenging cases, such as J0844$-$0132, we are able to accurately recover the host galaxy properties, as demonstrated in Figure \ref{['fig:sim_onecase']}.
• Figure 3: Same as Figure \ref{['fig:F356Wfit']} for the F150W filter. Note that for non-significant detections (i.e., those indicated by dashed-line elliptical apertures), the inferred host properties have larger uncertainties. In these cases, the orientation of the inferred host shape in F150W can even be inconsistent with that in F356W.
• Figure 4: Host-to-total flux ratio as a function of the quasar's absolute magnitude. The plot includes three luminous $z\sim6$ quasars reported in Yue2024. Our results demonstrate that the intrinsically low luminosity of our quasars leads to a significantly higher host-to-total flux ratio than found in luminous quasars. Note that the photometric measurements of the host galaxies in Yue2024 are based on a different methodology that results in very small statistical uncertainties, leading to exceptionally small error bars in their sample. The 3% threshold, which serves as one of our detection criteria for hosts, is indicated by the dashed line.
• Figure 5: This figure presents the SED inference for our host galaxies, derived using the gsf software. Red data points with error bars represent the inferred host galaxy fluxes across multiple bands. The gray-shaded region illustrates the 1-$\sigma$ range of SED template variations obtained through MCMC sampling. The black line denotes the median SED template. Blue diamonds indicate the flux predictions based on this median template for each observed band. We use eight-band photometry to infer the SED for J2236+0032 Onoue2024.