A Possible Shutting-Down Event of Mass Accretion in An Active Galactic Nucleus at z~1.8

TL;DR

This study investigates rare, large-amplitude fading events in active galactic nuclei at high redshift to understand turning-off of mass accretion. It leverages long-baseline photometry from SDSS and HSC to identify extreme fading quasars, using cross-survey selection and multi-epoch SED decomposition to separate evolving AGN emission from a constant host. The key finding is that J0218-0036 shows a dramatic decline by about a factor of $50$ in AGN luminosity between the early 2000s and 2023, with the Eddington ratio dropping from roughly $0.4$ to $0.008$, consistent with a substantial accretion-rate decrease rather than obscuration; by 2023 the host light dominates the rest-frame optical emission. This provides empirical evidence for shutting-down accretion events in SMBHs at $z\sim1.8$ and demonstrates the power of long-term, multi-epoch surveys to capture dramatic AGN state changes, informing SMBH growth and feedback models.

Abstract

We present the discovery of a large gradual apparent fading event in optical and near-infrared wavelengths in a quasar at z=1.767 by a factor of 20-30 (in optical) over a period of ~20 years in the observed frame. This pronounced fading trend in brightness was first identified by comparing the magnitudes measured in the Subaru/Hyper Suprime-Cam (HSC) images with those in the Sloan Digital Sky Survey (SDSS) images for ~3x10^4 quasars spectroscopically identified by SDSS. We performed follow-up observations, including optical imaging and spectroscopy as well as near-infrared imaging, with >4m-class telescopes such as Subaru, GTC, Keck, and SOAR telescopes. We combine these new data with the archival data to examine the variability behavior over ~20 years in detail and even the longer-term trend of the variability over ~70 years in the observed frame. We find that (i) the AGN component likely faded by a factor of ~50 from the early 2000s to 2023 and (ii) the observed brightness decline is best explained by a substantial decrease in accretion rate rather than time-varying line-of-sight dust obscuration. These findings are derived from multi-component (time-varying AGN + constant galaxy) spectral energy distribution fitting over multi-epochs, which is well consistent with the optical spectra. The Eddington ratio decreases by a factor of ~50, from ~0.4 to ~0.008 if we use the black hole mass measured with the SDSS spectrum, which could be highly uncertain because of the very large variability. The total brightness is dominated by the host galaxy in the rest-frame optical wavelength rather than the AGN as of 2023.