The cosmic X-ray background is the faint, diffuse glow that blankets the sky in X-rays, a celestial sunrise stitched together from the hearts of galaxies. Most of that glow comes from active galactic nuclei, the hungry cores of galaxies where matter rushes into supermassive black holes and their violent appetites fuel light across the spectrum. In the soft X-ray band, below about 10 keV, astrophysicists have learned to count up known AGN and model how much intervening material — gas and dust — sits between us and the bright center. But as we push toward higher energies, a stubborn excess appears, a signal that hints at a population of sources that are almost entirely hidden from view.
That hidden population is the target of a Clemson-led team, part of the CI-CTAGN project, which seeks to census Compton-thick AGN in the local universe. In a study led by Isaiah S. Cox and including Nuria Torres-Albà and Stefano Marchesi among others, researchers combined soft X-ray glimpses from Chandra with hard X-ray fingerprints from Swift-BAT to hunt for six nearby AGN that ROSAT couldn’t see in the soft band. They then tested two physically motivated models of the obscuring torus—borus02 and UXCLUMPY—to infer how much material lies along our line of sight and what the torus might actually look like. The big question they aimed to answer: how much light do these hidden giants contribute to the cosmic X-ray background, especially around the enigmatic Compton hump near 30 keV?
What are Compton-thick AGN and why they matter
Compton-thick (CT) AGN are the most shrouded versions of active galactic nuclei. When the line-of-sight hydrogen column density NH,los passes above roughly 1.5 × 10^24 cm−2, the surrounding matter becomes effectively opaque to X-rays. The primary X-ray light from the corona is heavily absorbed, so what we predominantly observe is light that has bounced off the obscuring material and been reprocessed—the reflection component. In the spectrum, that reprocessed light can create a pronounced peak called the Compton hump, sitting around 30 keV. In CT-AGN, the reflection can even outshine the direct emission at these energies, making them tricky to detect but incredibly informative about the hidden geometry around the black hole.
The practical importance of CT-AGN lies in their imprint on the cosmic X-ray background (CXB). The CXB is not a uniform glow; it encodes the collective history of black hole growth and obscuring gas across cosmic time. How many CT-AGN lurk in the local neighborhood, and how their obscuring material is arranged, strongly shapes the CXB’s high-energy tail. Early models suggested a potentially large CT fraction locally, but direct, hard X-ray detections have often fallen short of those predictions. That mismatch has left room for debate about how much reprocessed light CT-AGN contribute and what their torus looks like—clumpy clouds or a smoother doughnut, or something in between.
To tackle this, Cox and colleagues turned to two complementary torus models. borus02 envisions a fairly uniform density torus with a variable opening angle, allowing a range of covering factors. UXCLUMPY introduces a more mosaic-like torus, with clouds arranged in a probabilistic canopy. Each model encodes a different intuition about how gas and dust cradle the black hole, but both predict how X-rays should be absorbed and reflected. By fitting the same data with both models, the team could test the robustness of their NH,los measurements and gauge how much the inferred obscuration depended on assumed geometry. The result is a stronger, more credible census of CT-AGN in our cosmic backyard and a firmer anchor for CXB models.
How researchers hunted hidden giants
The six sources in this study were drawn from the Swift-BAT catalog, which surveys the sky in the hard X-ray window (15–150 keV) and is less biased against heavily obscured objects. Each one lacked a ROSAT soft X-ray counterpart, a signpost that the X-ray light from the nucleus might be heavily absorbed. The team then chased them with Chandra, obtaining snapshot observations of about 10 kiloseconds per target to gather clean soft X-ray spectra, and in three cases supplemented the data with XMM-Newton observations for greater soft X-ray coverage. The goal was to stitch together 1–150 keV spectra so that the true brightness of the AGN and the degree of obscuration could be disentangled with as much fidelity as current data allow.
Two statistical paths ran in parallel. First, a traditional Levenberg–Marquardt least-squares fit, which is fast and familiar but can get stuck in local minima if the model behaves nonlinearly. Second, a Bayesian approach using nested sampling to map out the full posterior distributions of NH,los and torus geometry parameters. In this Bayesian view, there isn’t a single “best” number; there is a distribution that captures what the data really say, including uncertainties and interdependencies among parameters. Reporting the mode (the most probable value) alongside a credible interval helps readers appreciate both the most likely answer and how confident we should be about it.
Why run both methods? Because the torus geometry—whether a smooth donut or a patchwork of clouds—can tilt the interpretation of the absorption signal. The researchers used two main torus models as their workhorses: borus02 and UXCLUMPY. borus02 parameterizes a uniform-density torus with a freely varying covering factor cf and an inclination angle, while UXCLUMPY allows a distribution of clouds whose combined absorption depends on how many clouds lie along the line of sight. The team also ran simpler variants (borus02* and MYTorus) as sanity checks when the data could not strongly constrain the more complex torus geometry. The processing was nontrivial; the team leveraged Clemson University’s Palmetto supercluster to run the Bayesian fits, which could take hours to days per source depending on model complexity.
Across all six sources, the NH,los measurements came out consistent between borus02 and UXCLUMPY, despite their different geometry assumptions. That convergence is reassuring: it suggests the soft-to-hard X-ray data are telling a robust story about how much material sits in the observer’s path to the black hole. In turn, those NH,los values underpin whether a source is truly Compton-thick or merely heavily obscured. The analysis also examined how often the torus geometry can be pinned down with snapshot data; not often, the authors note, because geometry-specific parameters (like the exact cloud distribution) usually require deeper, high-energy coverage from NuSTAR to constrain reliably.
Three of the six sources also had two soft X-ray observations, enabling a search for NH,los variability. The study introduced a practical metric, the probability of variation (Pvar), computed from how the NH,los posterior masses for the different epochs overlapped. The standout case was 2MASX J17253053–4510279, which showed strong hints of NH,los variability across epochs, even though both soft bands remained relatively unobscured. Another two sources showed only modest evidence for variability, while the rest appeared steadier. Yet the overall message remains clear: the X-ray sky is dynamic on human timescales, and that variability can be a crucial clue to the structure and evolution of the obscuring material around a growing black hole.
What the six sources reveal about the CXB and the torus
Putting the six AGN side by side yields a nuanced, data-driven picture of the local CT-AGN population. Two sources, 2MASX J17253053–4510279 and MCG +2-57-2, emerge as unobscured or only lightly obscured in the soft X-ray band, confirming that not all BAT-selected AGN are veiled. The other four are more intriguing: 2MFGC 9836 and IC 1141 carry obscuration consistent with Compton-thin absorption, while CGCG 1822.3+2053 and NGC 5759 sit in the heavier, more ambiguous end of the spectrum. Among these, NGC 5759 shines as the strongest Compton-thick candidate in the sample, with its NH,los posterior mass leaning firmly into the CT regime, especially in the Chandra data. CGCG 1822.3+2053 is a more tentative CT candidate, with a lower probability mass in the CT interval, but still a compelling case that warrants follow-up with NuSTAR to settle the question.
What do these six sources teach us beyond their individual classifications? First, the agreement between borus02 and UXCLUMPY NH,los values across both methods reinforces that the line-of-sight obscuration is a robust observable even when the torus geometry is not perfectly pinned down. That robustness is essential if we want to scale up from a handful of galaxies to a reliable census of CT-AGN in the local universe. Second, the study highlights how multi-instrument, broad-band spectroscopy — soft X-ray data from Chandra (and XMM-Newton when available) tied to hard X-ray light from Swift-BAT — is a powerful way to break degeneracies that plague fitting X-ray spectra of obscured sources. Finally, while NH,los is becoming well constrained, the torus’ geometry (how much sky it covers, how clumpy it is) remains the harder nut to crack without deeper NuSTAR data, reminding us that black holes live in three-dimensional shadows, not two-dimensional silhouettes.
From a cosmological perspective, each CT-AGN we confirm or constrain tightens the grid on the CXB’s high-energy tail. CT-AGN are the hidden engines of the hard X-ray sky; mapping them in the local volume helps anchor population synthesis models that extrapolate to higher redshifts. The authors’ approach—combining Swift-BAT’s hard X-ray view with Chandra/XMM-Newton’s soft X-ray insight and testing two distinct torus geometries—provides a blueprint for building a more complete, less biased portrait of the local CT-AGN population and, by extension, a clearer map of the CXB’s origins. The upshot is not just cataloging six sources; it’s refining a method by which the universe’s most enshrouded engines can be counted, characterized, and finally understood in the grand story of black hole growth.
The study is anchored in Clemson University’s Department of Physics and Astronomy, with collaborators from INAF in Bologna, Caltech, and the University of Illinois, and the lead author is Isaiah S. Cox. The work represents a measured step forward in the long, ongoing project to illuminate the CXB not by forcing light through a veil, but by mapping the veil itself—its thickness, its texture, and its temperament over time—and learning how those properties sculpt the glow we see across the cosmos.
