🔭 astrophysics

The incidence of eROSITA X-ray AGN in the local Universe: from dwarf to massive galaxies

By combining the eROSITA All Sky Survey with multi-wavelength catalogues, this study establishes a cumulative ~1% AGN fraction in low-mass local galaxies, providing a firm lower limit on black hole occupation and revealing a nuanced, mass-dependent picture of AGN growth that challenges current modeling.

Original authors: Z. Igo, A. Merloni, A. Georgakakis, J. Buchner, R. Arcodia, M. Salvato, J. Aird, K. Nandra, B. Trakhtenbrot, P. G. Boorman, J. Comparat, G. Lamer, B. Laloux, M. Kluge, W. Roster, E. Bulbul, F. Balzer

Published 2026-07-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Z. Igo, A. Merloni, A. Georgakakis, J. Buchner, R. Arcodia, M. Salvato, J. Aird, K. Nandra, B. Trakhtenbrot, P. G. Boorman, J. Comparat, G. Lamer, B. Laloux, M. Kluge, W. Roster, E. Bulbul, F. Balzer, T. Dwelly, W. N. Brandt, R. Seppi, S. Morrison, E. Kyritsis, J. Gelfand, S. F. Anderson, D. P. Schneider

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe as a giant, cosmic neighborhood. For a long time, astronomers thought that the "superstars" of this neighborhood—massive galaxies with giant black holes in their centers—were the only ones with active, hungry monsters. The tiny, dwarf galaxies were thought to be quiet, empty houses where the black holes either didn't exist or were fast asleep.

But a new study using a giant space telescope called eROSITA has come along to check the mailboxes of these tiny houses, and the results are shaking up the neighborhood watch.

The Great Cosmic Census
The researchers didn't just peek at a few houses; they took a massive census of about 5.35 million galaxies in our local cosmic neighborhood (specifically, those with a redshift between 0.03 and 0.2). They were looking for X-rays, which are like the cosmic equivalent of a "smoke signal" from a black hole that is actively eating gas.

Here is the tricky part: The telescope's view is a bit fuzzy, like looking through a foggy window. When they first looked at the tiny dwarf galaxies, they thought they saw smoke signals in 4,121 of them. But after a rigorous "clean-up" process—checking for false alarms, overlapping signals from neighbors, and random cosmic static—they realized that most of those signals were just noise. In fact, they had to throw out about 79% of the initial low-mass detections as false alarms!

The Real Discovery
After the dust settled, they found 874 genuine X-ray active black holes in dwarf galaxies. This is a big deal because it's the largest list of its kind ever made. Some of these tiny black holes are surprisingly loud, blasting out X-rays with a power of more than 10⁴³ erg s⁻¹. That's like a tiny firecracker suddenly screaming as loud as a jet engine.

The "Appetite" of Black Holes
The team didn't just count them; they measured how hungry these black holes are. They used a metric called the "specific accretion rate" (λSAR), which is basically a measure of how much food the black hole is eating relative to the size of its host galaxy.

They found that the "hunger" of black holes isn't the same for everyone.

The Break: They noticed a sharp "break" in the data at high hunger levels (around λSAR ≥ 10⁻²). It's as if the black holes hit a speed limit. When they try to eat too fast, something stops them—likely a self-regulating mechanism where the radiation pressure blows the food away.
The Mass Difference: Interestingly, the biggest galaxies seem to have a harder time reaching these super-high hunger levels compared to medium-sized galaxies. It's like a giant elephant struggling to eat as fast as a medium-sized dog.
The Low-Mass Mystery: In the smallest galaxies, the number of active black holes drops off significantly. The study suggests that the "efficiency" of feeding these black holes is lower in the tiniest galaxies compared to those with a stellar mass around log M∗/M⊙ ∼ 10 – 10.5.

What This Means for the Neighborhood
The study calculates that for the smallest galaxies (those with a stellar mass log M∗/M⊙ ≤ 10), only about 1% (or less) of them have a black hole that is currently active and eating at a significant rate.

This doesn't mean the other 99% don't have black holes; it just means they are quiet. The authors are careful to say this is a "firm lower limit." It's like saying, "We found 1% of the houses with the lights on." It doesn't prove the other houses are empty; they might just have the lights off. However, it does prove that active black holes do exist in these tiny galaxies, and they can be surprisingly powerful.

What the Paper Rules Out
The paper explicitly argues against the idea that the distribution of black hole hunger is the same for all galaxy sizes. Previous studies suggested that a black hole's eating habits were independent of its host galaxy's size, but this new data shows that size does matter. The "rules" for how black holes grow change depending on whether they live in a mansion or a shack.

The paper also rules out the idea that the initial flood of signals from the tiny galaxies was real. The "noise" was so high that without their strict cleaning process, the results would have been completely wrong.

How Sure Are They?
The authors are very confident in the 874 active black holes they identified because they used a sophisticated statistical method (a Bayesian framework) that accounts for background noise and false alarms. They are also confident that the "break" in the hunger curve is real, thanks to the sheer size of their sample.

However, they are less sure about what happens at the very lowest levels of hunger. Because the telescope isn't sensitive enough to see the faintest "whispers" of X-rays, they can't perfectly map out the quietest black holes. They suggest that the true number of black holes might be higher, but the ones they can see are definitely there.

The Bottom Line
This study is like finding out that even the smallest, quietest houses in the neighborhood have a secret, active kitchen. It challenges the old idea that only the big houses have the resources to feed a black hole. While we still don't know exactly how many black holes are hiding in the dark, we now know for sure that the tiny ones are there, and when they do wake up, they can be surprisingly loud. This helps scientists understand how the very first black holes in the universe might have started, using our local neighborhood as a time machine.

Technical Summary: The Incidence of eROSITA X-ray AGN in the Local Universe

Problem Statement
The incidence of supermassive black holes (SMBHs) in the centers of low-mass galaxies (dwarfs, $M_* \le 10^{10} M_\odot$ ) remains a critical open question in astrophysics, with implications for understanding black hole seeding mechanisms in the early Universe. While SMBHs are ubiquitous in massive galaxies, their occupation fraction (BHOF) in dwarfs is debated, with different seeding models (light vs. heavy seeds) predicting vastly different occupation rates. Observationally, determining the "active fraction" (the subset of galaxies hosting an accreting AGN) in low-mass systems has been hindered by small sample sizes and selection biases in previous studies. Furthermore, distinguishing nuclear AGN emission from other galactic X-ray processes (e.g., X-ray binaries, hot gas) in low-mass galaxies is challenging due to the limited spatial resolution of X-ray telescopes and the faintness of the sources.

Methodology
This study constructs the largest statistical sample of X-ray AGN in the local Universe ( $0.03 < z < 0.2$ ) by combining the four-pass eROSITA All-Sky Survey (eRASS:4) with the 10th Data Release of the DESI Legacy Imaging Survey (LS10).

Parent Sample Construction: A parent sample of $\sim 5.35$ million galaxies was selected from LS10 based on $z$ -band flux ( $12 < z_{mag} \le 20$ ) and redshift. The sample was rigorously cleaned to remove contaminants (e.g., stellar sources, Gaia duplicates, sources near bright foreground galaxies or X-ray clusters).
Physical Property Estimation: Stellar masses ( $M_*$ ) and star formation rates (SFR) were derived via Spectral Energy Distribution (SED) fitting. A machine-learning classifier (XGB) identified sources where standard SED fitting (LePHARE) might be biased by AGN emission; these were re-analyzed using the Bayesian code GRAHSP, which includes AGN and mid-IR dust components.
X-ray Detection and Cleaning: X-ray fluxes were measured via aperture photometry centered on optical galaxy coordinates using the apetool task. To address the high rate of spurious associations in low-mass galaxies (due to the large eROSITA Point Spread Function), a multi-step validation process was employed:
- Cross-matching with eRASS:4 source catalogs.
- Statistical validation using NWAY (counterpart probability).
- Visual inspection of low-mass candidates.
- Creation of a "spurious emission" model using offset apertures to empirically estimate the background contamination distribution.
Bayesian Hierarchical Modeling: The authors employed a Bayesian framework to infer the intrinsic distribution of specific accretion rates, $p(\log \lambda_{SAR} | M_*, z)$ $p (lo g λ_{S A R} ∣ M_{*}, z)$ , where $\lambda_{SAR} \propto L_X / M_*$ $λ_{S A R} \propto L_{X} / M_{*}$ . Two models were tested:
- Model A: Accounts for AGN, galactic processes (XRBs/hot gas), and background.
- Model B: Includes an additional component for spurious detections derived from the offset aperture analysis.
  The models marginalize over uncertainties in black hole mass (via the $M_{BH}-M_*$ relation) and galactic X-ray emission.

Key Contributions

Sample Size: The study identifies 874 X-ray AGN in low-mass galaxies ( $M_* \le 10^{10} M_\odot$ ) and 12,618 in high-mass galaxies, representing the largest statistical sample of its kind in the local Universe.
Rigorous Cleaning: The work highlights that $\sim 79\%$ of initial X-ray detections in low-mass galaxies are spurious or misidentified counterparts, necessitating the extensive cleaning and validation procedures developed here.
Unified Framework: The application of a Bayesian hierarchical model that simultaneously utilizes X-ray counts from both detected and undetected sources allows for the derivation of the specific accretion rate distribution across a wide dynamic range, including regimes where direct detection is impossible.

Results

Accretion Rate Distribution: The specific accretion rate distribution, $p(\log \lambda_{SAR} | M_*, z)$ $p (lo g λ_{S A R} ∣ M_{*}, z)$ , exhibits a complex, mass-dependent structure.
- A distinct break is observed at high $\lambda_{SAR}$ ( $\sim 10^{-2} - 10^{-1}$ ), suggesting a transition to Eddington-limited, self-regulated black hole growth.
- The break position shifts to lower $\lambda_{SAR}$ values for higher stellar masses, indicating that massive galaxies are less likely to reach high accretion rates compared to lower-mass systems.
- In the low-mass regime ( $\log M_*/M_\odot \le 9.5$ ), the normalization of the distribution is lower than extrapolations from high-mass galaxy studies (e.g., Aird et al. 2012).
Cumulative AGN Fraction:
- For $\lambda_{SAR} \ge 10^{-3}$ , the cumulative AGN fraction in low-mass galaxies is constrained to $\lesssim 1\%$ . This provides a firm lower limit on the black hole occupation fraction in this mass regime.
- The fraction of highly accreting AGN ( $\lambda_{SAR} \ge 10^{-2}$ ) peaks around $\log M_*/M_\odot \sim 10 - 10.5$ , showing a suppression of fueling efficiency at both lower and higher masses.
- The study notes that these fractions are lower limits for unobscured sources; correcting for obscuration (assuming mass-independent obscuration trends) could increase the fractions by factors of $\sim 2-4$ .
High-Luminosity Low-Mass AGN: Several low-mass galaxies host AGN with $L_{2-10 \text{ keV}} > 10^{43} \text{ erg s}^{-1}$ , challenging simple accretion models that struggle to fuel low-mass black holes to such luminosities.

Significance and Claims
The paper claims to provide the most robust statistical characterization of X-ray AGN incidence in low-mass galaxies to date. By revealing a mass-dependent structure in the specific accretion rate distribution and a break at high accretion rates, the authors argue that AGN growth is not a simple, mass-invariant stochastic process. The finding of a $\sim 1\%$ active fraction in low-mass galaxies sets a strict lower bound on the black hole occupation fraction, though the authors caution that degeneracies between occupation and duty cycle prevent a definitive measurement of the total BHOF from X-ray data alone.

The results suggest that local low-mass galaxies may not be perfect analogues for high-redshift primordial galaxies if feedback mechanisms (which can regulate growth) are already active in these systems. The presence of highly accreting AGN in dwarfs implies that AGN feedback could play a role in low-mass galaxy evolution comparable to or exceeding stellar feedback, a factor that must be incorporated into future simulations of galaxy formation and black hole seeding. The study emphasizes that future constraints on seeding models will require deeper X-ray surveys (e.g., Athena) and multi-wavelength variability studies (e.g., LSST) to probe the fainter, lower-accretion end of the population.

More like this