Mid-infrared Variability-based AGN Selection using the Multi-epoch Photometric Data from WISE

Abstract

We assess the systematics and efficiency of an AGN selection method based on mid-infrared (MIR) variability. To this end, we utilize various types of active and inactive galaxies from the Sloan Digital Sky Survey, matching them with multi-epoch photometric data from the NEOWISE mission. Using W1 and W2 band light curves with a $\sim10$-year baseline, we find that combining the likelihood of deviation from non-variability with the correlation coefficient between the W1 and W2 bands reliably identifies AGNs. Specifically, this MIR-based method recovers $\sim 28.2\%$ of optically selected AGNs. Applying the same technique to inactive galaxies, we identify AGN candidates at fractions ranging from $0.4$ to $11.8\%$, indicating that MIR variability allows us to detect AGN candidates even in optically inactive hosts. While some variable sources exhibit transient-like light curves, possibly originating from tidal disruption events or supernovae, their contribution to the total variable population is less than a few percent, indicating a minimal impact on our results. Across all subsamples, the AGN fraction marginally increases with star formation activity, implying coordinated evolution between central black hole growth and star formation. Finally, the AGN fraction inferred from our method drops dramatically in classical LINERs, consistent with their low accretion rates and absence of a dusty torus.

Figures (13)

• Figure 1: Two examples of light curves from our AGN sample. The faint dots denote the original data prior to binning, while the large circles represent the binned data. Due to the artificial offset between AllWISE and NEOWISE, we show only the photometric data from NEOWISE. The top and bottom panels correspond to the W1 and W2 magnitudes, respectively. The left panel shows an example that fulfills the Group 1 criteria but has $r < 0.75$, whereas the right panel shows an example that satisfies the Group 2 criteria.
• Figure 2: The BPT and VO diagrams for the optically selected AGN sample are shown. The fraction of MIR variability–selected AGNs ($f_{\rm var}$) in each bin is indicated by the color maps. The original AGNs were classified as sources above the solid line adopted from kewley_2001. SF galaxies were selected below the dashed line kauffmann_2003, while sources lying between the dashed and solid lines were classified as composites. From top to bottom, MIR-based AGN classifications are presented in different categories. $f_{\rm var}$ is shown only in bins with more than 10 sources.
• Figure 3: Same as Figure 2, except that $f_{\rm var}$ is shown for SF galaxies and composites. The subgroups were originally categorized in the BPT diagram (i.e., [O3]/H$\beta$ vs. [N2]/H$\alpha$)
• Figure 4: Distributions of MIR variability–based AGN fractions in the W1–W2 versus W2–W3 color diagram. The contours represent the density distribution of the parent sample, while the color map shows the fraction of AGN sources in each bin, categorized according to the three different criteria. The AGN fraction is only demonstrated in bins containing more than 10 sources. The thick dashed lines show the AGN wedge adopted from mateos_2012.
• Figure 5: Same as Figure 4, for different subsamples.