The Suzaku View of the Swift/BAT AGNs (II): Time Variability and Spectra of Five "Hidden" AGNs

The fraction of Compton thick sources is one of the main uncertainties left in understanding the AGN population. The Swift Burst Alert Telescope (BAT) all-sky survey, for the first time gives us an unbiased sample of AGN for all but the most heavily absorbed sources (log NH > 25). Still, the BAT spectra (14 - 195 keV) are time-averaged over months of observations and therefore hard to compare with softer spectra from the Swift XRT or other missions. This makes it difficult to distinguish between Compton-thin and Compton-thick models. With Suzaku, we have obtained simultaneous hard (> 15 keV) and soft (0.3 - 10 keV) X-ray spectra for 5 Compton-thick candidate sources. We report on the spectra and a comparison with the BAT and earlier XMM observations. Based on both flux variability and spectral shape, we conclude that these hidden sources are not Compton-thick. We also report on a possible correlation between excess variance and Swift BAT luminosity from the 16 d binned light curves, which holds true for a sample of both absorbed (4 sources), unabsorbed (8 sources), and Compton thick (Circinus) AGN, but is weak in the 64 day binned BAT light curves.

💡 Research Summary

The paper addresses one of the lingering uncertainties in active‑galactic‑nucleus (AGN) demographics: the true fraction of Compton‑thick (CT) sources (NH > 1.5 × 10²⁴ cm⁻²). The Swift Burst Alert Telescope (BAT) all‑sky survey provides an unprecedented, nearly unbiased AGN sample in the 14–195 keV band, but its spectra are averaged over months, smoothing out short‑term variability. This time‑averaging hampers the discrimination between CT and Compton‑thin (CT‑thin) models, especially for sources that appear “hidden” in softer X‑ray observations.

To overcome this limitation, the authors selected five BAT‑detected “hidden” AGN that were previously flagged as CT candidates based on their hard‑band fluxes and lack of soft X‑ray detections. They obtained simultaneous Suzaku observations, covering the soft X‑ray range (0.3–10 keV) with the XIS detectors and the hard X‑ray range (>15 keV) with the HXD‑PIN. By acquiring truly simultaneous spectra, the study eliminates the systematic bias introduced by comparing a month‑averaged BAT spectrum with a snapshot soft‑X‑ray observation.

Data reduction followed standard HEASOFT procedures: XIS events were cleaned, HXD‑PIN background was modeled, and cross‑normalization constants between instruments were applied. The authors also extracted BAT light curves binned at 16 days and 64 days, calculating the excess variance (σ²_rms) for each bin to quantify variability.

Spectral fitting employed an absorbed power‑law model (phabs*zpowerlw) supplemented by a reflection component (pexrav) and a narrow Fe Kα line. For all five objects, the best‑fit column densities lie in the range NH ≈ (1–3) × 10²³ cm⁻², well below the CT threshold. The reflection scaling factors are modest (R ≈ 0.3–0.5), and no strong Compton‑hump is evident. Moreover, the 2–10 keV fluxes, when compared to the BAT 14–195 keV fluxes, are inconsistent with the severe suppression expected for CT sources. These spectral results alone already argue that the objects are CT‑thin, despite their “hidden” classification in earlier surveys.

Variability analysis provides an independent line of evidence. All five sources show measurable excess variance on the 16‑day timescale (σ²_rms ≈ 0.01–0.03), indicating intrinsic flux changes on weeks‑long intervals. The authors find a positive correlation between the excess variance and the BAT luminosity derived from the 16‑day binned light curves. To test the robustness of this trend, they extend the analysis to a larger set comprising four absorbed AGN, eight unabsorbed AGN, and the well‑studied CT galaxy Circinus. The correlation persists across this heterogeneous sample, but it weakens dramatically when the BAT light curves are rebinned to 64 days, suggesting that the variability‑luminosity link is most apparent on relatively short (∼weeks) timescales.

The discussion emphasizes two key implications. First, reliance on BAT‑averaged spectra can misclassify CT‑thin AGN as CT because variability can temporarily boost the hard‑band flux, mimicking the flat, reflection‑dominated spectra of true CT objects. Simultaneous soft‑hard coverage, as provided by Suzaku, is essential for accurate column‑density determination. Second, the observed excess‑variance–BAT‑luminosity correlation may serve as a diagnostic of the geometry and clumpiness of the obscuring torus. Sources with higher variability likely have a more porous or dynamically active absorber, allowing the intrinsic continuum to leak through and produce stronger hard‑band emission.

In conclusion, the five “hidden” AGN studied here are not Compton‑thick; they are heavily absorbed but CT‑thin. The work demonstrates that time‑resolved, broadband X‑ray spectroscopy is crucial for disentangling absorption effects from intrinsic spectral shape. Moreover, the tentative variability‑luminosity relation opens a new avenue for probing the structure of the circumnuclear material in AGN. Future campaigns that combine long‑term monitoring (e.g., with Swift/BAT) and simultaneous broadband observations (e.g., with NuSTAR, XMM‑Newton, or future missions) will refine the CT fraction and improve our understanding of how supermassive black holes grow behind thick veils of gas and dust.