The Dichotomy in the Nuclear and Host Galaxy Properties of High-redshift Quasars

TL;DR

High-$z$ quasars illuminate rapid black hole growth and host-galaxy coevolution, but disentangling AGN and host light remains challenging. The authors present GalfitS, a multiband forward-modeling tool, to perform self-consistent AGN–host decomposition on a sample of $31$ quasars at $4 \lesssim z \lesssim 7$ using JWST/NIRCam imaging from multiple surveys. They report that quasar hosts are typically compact by $\sim0.3$ dex relative to star-forming galaxies of similar mass, and uncover a dichotomy: luminous blue quasars with $L_{5100}\gtrsim 10^{45}\ \mathrm{erg\,s^{-1}}$ reside in bulge-dominated hosts ($n \approx 5$) with a narrow UV slope $\beta_{\rm UV} \approx -1.4$, while fainter red quasars live in disk-like hosts ($n \approx 1$) with a broad $\beta_{\rm UV}$ range from $-2$ to $4$. The inferred $M_{\rm BH}/M_\ast$ ratios differ markedly, with high-luminosity sources at $M_{\rm BH}/M_\ast = 1.2\%$ and lower-luminosity sources at $4.7\%$, placing the ensemble roughly $0.6$ dex above the local $M_{\rm BH}-M_\ast$ relation, consistent with rapid BH growth in gas-rich environments and feedback becoming effective after substantial gas depletion. Overall, the results suggest a fundamental dichotomy in the coevolution of supermassive black holes and their hosts in the early Universe, potentially driven by intrinsic spectral differences or dust extinction.

Abstract

The early growth of high-redshift quasars and their host galaxies raises critical questions about their cosmic evolution. We exploit the angular resolution and sensitivity of NIRCam to investigate the host galaxies of 31 quasars at $4\lesssim z\lesssim7$ drawn from multiple JWST surveys. Using a new multi-band forward-modeling code (\textsc{GalfitS}) that incorporates physically motivated priors, we securely detect and quantify the host emission in 30 objects, while simultaneously characterizing the nuclear spectral energy distribution. The host galaxies of high-redshift quasars are $\sim 0.3$~dex more compact than star-forming galaxies of comparable mass. A striking dichotomy emerges: luminous ``blue'' quasars ($L_{5100}\gtrsim10^{45}\,{\rm erg\,s^{-1}}$) reside in bulge-dominated galaxies ($n \approx 5$) and exhibit a narrow range of ultraviolet nuclear slopes (median $β_{\rm UV} \approx -1.4$), while fainter ``red'' quasars inhabit disk-like hosts ($n\approx 1$) and display a broad range of slopes ($β_{\rm UV}\approx-2$ to 4). These two populations differ markedly in their black hole-to-stellar mass ratios, with high-luminosity quasars showing $M_{\mathrm{BH}}/M_\ast = 1.2\%$ compared to $4.7\%$ for lower luminosity sources, placing them collectively $\sim$0.6~dex above the local $M_{\mathrm{BH}}-M_\ast$ relation. This offset likely reflects rapid black hole growth in early gas-rich environments, where feedback from the active galactic nucleus becomes effective only after substantial gas depletion. Our findings suggest that the observed dichotomy, whether due to intrinsic spectral differences or dust extinction, fundamentally shapes the coevolution of supermassive black holes and their host galaxies in the early Universe.

Figures (1)

• Figure 1: Top row: examples of PSF models constructed for the seven NIRCam bands from CEERS pointing 1 data. Bottom row: median surface brightness profiles for fitted PSF models across all target fields (black lines), with the shaded grey region indicating one standard deviation. Median surface brightness profiles for high-luminosity and low-luminosity quasars are shown as blue and magenta points, respectively, with error bars representing one standard deviation. All profiles are normalized by their total flux.