AGN ruled out as the dominant source of cosmic reionization

Abstract

Cosmic reionization represents the latest phase transition in the Universe, when the Lyman continuum (LyC) photons turned the intergalactic medium (IGM) from neutral to highly ionized. It has long been debated whether galaxies or active galactic nuclei (AGNs) are the major source of LyC photons responsible for reionization. Previous observations slightly favored galaxies as the major ionizing source. However, the James Webb Space Telescope (JWST) recently discovered an unexpectedly high density of AGN candidates at high redshift, which has largely enhanced the influence of AGNs. Here we derive a definitive upper bound on the AGN contribution to reionization using the latest JWST data, and conclusively rule out AGNs as the dominant ionizing source during the peak epoch of reionization (EoR). We build a sample of galaxies and AGNs in a specific redshift range $7.15 \leq z \leq 7.75$ with a high completeness. Each object is then decomposed into a point-source component and an extended component in their rest-frame far-UV JWST images. We assume all point-source components are AGNs. Our fiducial AGN sample reaches an unprecedentedly low luminosity of $M_{\rm UV} \approx -15$ mag. Based on this sample, AGNs can contribute at most one third of the LyC photons budget required at $z\sim7.5$. Our result implies that galaxies dominate the ionizing source during the EoR.

Figures (4)

• Figure 1: Color selection of the high-redshift sample. The three spectra represent a model spectrum of quasars (AGNs) that has been redshifted to $z =$ 7.15, 7.40, and 7.75, respectively (Methods). The line colors of the spectra depend on redshift. The two black filter transmission curves represent JWST F090W and F115W filters. The inset shows how the F090W–F115W color changes with redshift. A galaxy component would slightly increase the F090W–F115W color. We use $\rm F090W-F115W>2$ to efficiently remove low-redshift objects.
• Figure 2: Two examples of JWST image decomposition in the F115W and F150W bands. The PSF fractions in the two cases are $\sim 40\%$ and $\sim 70\%$, respectively. The size of each image is 21 pixels (0.63 arc-seconds) on a side, centered on the flux peaks of the targets. The five columns from left to right display the following information: $\left(1\right)$ the observed JWST image; $\left(2\right)$ the best-fit model consisting of a PSF plus a Sérsic host galaxy; $\left(3\right)$ the PSF-subtracted image; $\left(4\right)$ the fitting residual image divided by the error image; $\left(5\right)$ the surface brightness profile. The vertical bars show the $1\sigma$ errors. The color maps reflect the flux level in the unit of counts per second on a logarithmic scale, as shown by the attached color bars. This image decomposition procedure works well for our sample.
• Figure 3: Distribution of the measured PSF fractions. The histograms show the PSF fractions for our sample of 155 high-redshift objects in F115W and F150W. The shaded region represents objects with PSF fractions below $20\%$, and they are excluded from our final sample. The two distributions are generally consistent.
• Figure 4: UV LF and the upper limit of the AGN contribution to reionization.a, UV LF at redshift $z=7.15-7.75$. The red circles represent our binned LF. Its horizontal bars indicate the luminosity ranges covered by the bins and the vertical bars show the $1\sigma$ errors of the LF. The solid red line with the shaded region represents our best-fit DPL LF with its 1 $\sigma$ region. The light purple dot-dashed line represents a galaxy LF at $z=7.5-8.5$ from ref. adams2024. The black dashed line represents the bright quasar LF (QLF) at $z\sim7$ from ref.matsuoka2023. The filled orange squares represent the upper limit of QSO LF at $z\sim6.2$ from ref.jiang2022 based on the observational data before JWST. The light blue open circles represent the LRDs LF at $z=6.5-8.5$ from ref.kokorev2024. The light green open circles represent the LF of red AGNs at $z=7-8$ from ref.greene2024. b, Upper limit of the AGN contribution to reionization. The solid red curve with its $\pm1\sigma$ uncertainty region represents the cumulative fraction ($f_{\mathrm{AGN}}$) of the maximum AGN emissivity to the total photon emissivity required to ionize the Universe. They are computed using the above LF. This result excludes the AGN-dominant scenario for cosmic reionization.