New Insight from the James Webb Space Telescope on Variable Active Galactic Nuclei

Abstract

Variability detected in galaxies is usually attributed to their active galactic nuclei (AGNs). While all AGNs are intrinsically variable, the AGN unification model predicts that type~2 AGNs rarely vary because their engines are blocked by dust tori. Previous UV-to-near-IR variability studies largely support this expectation. Here, we present a variability study by James Webb Space Telescope (JWST) that reveals a more subtle picture. Using NIRCam imaging data from three surveys over $\sim$140~arcmin$^2$ in the COSMOS field, we found 117 galaxies with $\geq 4$$σ$ variability in the F356W band across $\sim$2-year baseline. Cross-matching with the existing JWST spectroscopic data, we identified five of them at $z=0.19$--3.69 (F356W corresponding to rest-frame $λ\approx0.76-2.97$~$μ$m), which were all coincidentally observed by a NIRSpec program almost contemporaneously with the last imaging epoch. One additional variable was identified at $z=0.90$ using the archival Keck telescope data. These six objects form our spectroscopic subsample. Interestingly, two reside in close-pair environments, while two others form a close pair themselves. Most of their light curves can hardly be explained by nuclear transients, and AGN variability is a more plausible cause. However, among these six objects, (1) only one shows broad Bracket and Pfund series permitted lines ($Δv > 1000$~km~s$^{-1}$) indicative of a type~1 AGN; (2) two show narrow permitted lines (H$α$ and/or He~I$\lambda10830$) consistent with type~2 AGNs, with another one likely type~2 based on the host galaxy properties; and (3) two others, which form a pair, show no emission lines. Our results add more challenges to the unification model.

Figures (14)

• Figure 1: F356W magnitude differences between the C3D and PRM epochs versus C3D F356W magnitude. The small back dots are all sources in the field. The blue dashed curves represent the 4$\sigma$ statistical boundary of the distribution. The large gray, filled circles indicate the 117 objects retained in the variable sample. There are more brightening variables than fading ones because the C3D images were used for detections. The six variables that form the spectroscopic sample are shown as the red triangles.
• Figure 2: Summary figure for C3DvarN_0173. The first row shows its image stamps (36 on a side; North is up and East is to left) in the reference epoch (PRM.1) in all bands and their color composite. The second row shows the image stamps (same size and orientation as above) and their color composite in the discovery epoch (C3D.1) in the left panels, the difference images between the discovery and reference epochs in F115W, F200W, and F356W (C3D.1$-$PRM.1) in the middle panels, and the NIRSpec MSA slit placement superposed on the color composite in the right panel. The third row shows the SEDs of the variable in all epochs (left) and the light curves in the bands that have more than one epoch of photometry (right). The fourth row displays the NIRSpec 2D and 1D spectra. The traces of the variable and the neighbor are labeled in the 2D spectrum. The detected emission lines are marked on the 1D spectra for both.
• Figure 3: Similar to Figure \ref{['fig:n0173']} but for C3DvarS_0007.
• Figure 4: Similar to Figure \ref{['fig:n0173']} but for C3DvarS_0722. This object does not have any emission lines detected in the spectrum. However, there is a strong absorption feature at $\sim$1.5 $\mu$m, which we tentatively identify as the 4000 Å break. With this identification, the source is at $z=2.8$.
• Figure 5: Similar to Figure \ref{['fig:s0722']} but for C3DvarS_0723. Its redshift is also based on the absorption feature identified as the 4000Å break.