A survey of nine nearby galaxies has found that actively growing supermassive black holes may build stellar nurseries rather than destroy them — a finding that directly challenges the foundational assumption baked into every major galaxy-formation simulation currently in use. The study is linked to a CfA press release describing the pattern’s consistency.

Published September 14 in The Astrophysical Journal, the study “Common Excitation Patterns of Star Formation, Active Galactic Nuclei, and Shocks in Seyfert Galaxies” — led by Peixin Zhu, a graduate student and astronomer at the Center for Astrophysics | Harvard & Smithsonian — found that in each of the nine galaxies studied, the active black hole at the center is associated with a ring or arc of newly forming stars, not a region of suppressed star formation. The paper is available as arXiv preprint 2609.15913.

“We’re seeing that black holes are not just consuming material at the centers of galaxies, but they’re actively reshaping their surroundings,” said Lisa Kewley, an astrophysicist and director of the CfA who served as Zhu’s advisor and co-investigator on the study. “This work helps us understand a complex feedback cycle that plays an important role in galaxy evolution.” The finding’s consistency across all nine target galaxies — repeated not just in one outlier system but as a pattern — is what makes it editorially significant. A pattern observed in nine galaxies carries statistical weight that a single case study cannot.

What Galaxy Simulations Have Been Getting Wrong

To understand why this matters, it helps to know what the dominant scientific story has been. Cosmologists studying how galaxies form and grow have long relied on the concept of AGN feedback — the idea that when a galaxy’s central supermassive black hole becomes active, accreting nearby gas and dust, it releases energy in the form of jets and radiation-driven winds. That energy, the standard model predicts, heats and expels the surrounding molecular gas, cutting off the raw material that would otherwise collapse to form new stars. The galaxy falls quiet. This is called negative AGN feedback, and it is currently implemented as the primary (or only) AGN mechanism in the most widely used cosmological simulations — IllustrisTNG, EAGLE, and SIMBA.

The reason negative feedback was built into these simulations in the first place is pragmatic: something had to explain why the most massive galaxies in the universe stopped forming stars while lower-mass galaxies kept going. AGN feedback was the most plausible candidate. But “plausible” and “dominant” are not the same as “universally correct,” and direct observational confirmation at the scale of individual galaxies has proven elusive.

The new CfA study does not find negative feedback to be irrelevant. It finds something more nuanced and more difficult for simulation teams to simply accommodate with a parameter tweak: that the positive form of AGN feedback — in which energy outflows from the black hole compress surrounding gas rather than expel it, triggering new star formation rather than shutting it down — appears to be geometrically regular and statistically common across a representative sample of nearby active galaxies. These are not exotic extreme systems; they are ordinary Seyfert galaxies, a class that makes up a significant fraction of all spiral galaxies in the observable universe.

How the Team Made the Discovery: The New 3D Diagnostic

The technical breakthrough that made this finding possible is a new method for reading the light from galaxies. To see why it matters, consider the tool it improved upon.

Since 1981, the dominant approach to classifying what is happening inside a galaxy has been the BPT diagram, named for the astronomers Baldwin, Phillips, and Terlevich who developed it. A BPT diagram plots two pairs of optical emission-line ratios — most commonly the ratio of oxygen-III emission to hydrogen-beta emission, against the ratio of nitrogen-II emission to hydrogen-alpha emission — and uses where a galaxy’s spaxels fall on the resulting 2D plane to classify the dominant ionization source: star formation, AGN, or a composite of both. The approach has been foundational to observational astrophysics for four decades.

But the BPT diagram has a critical limitation: it cannot reliably separate AGN photoionization from fast shock excitation. When a black hole jet or wind slams into the surrounding interstellar gas at high speed, that collision produces shocks — violent compressions of the gas that heat it and generate their own emission-line signature. The problem is that this shock signature closely resembles the AGN signature on the BPT diagram. A 2D plot cannot disentangle them.

Zhu and her colleagues, including Kewley and Ralph Sutherland of the Australian National University — who provided the state-of-the-art theoretical shock and photoionization models central to the analysis — developed a solution: a three-dimensional diagnostic diagram that adds a third emission-line ratio to the BPT’s two. This third axis specifically separates fast shocks from AGN photoionization, enabling a pixel-by-pixel decomposition of each galaxy image into three distinct contribution maps: one for star formation, one for AGN radiation, and one for shock excitation.

“Once we resolved them, we could see that they not only accrete things, but they also eject things,” Zhu said. “The injection and accretion are linked with each other.”

The team applied this 3D method to data from the Multi Unit Spectroscopic Explorer (MUSE), the integral field unit spectrograph installed on the European Southern Observatory’s Very Large Telescope (VLT) in Paranal, Chile. MUSE is the most powerful wide-field optical IFU in existence: in Wide Field Mode, it covers a one-arcminute-square field of view at 0.2 arcsecond spatial resolution, splitting that field into 24 channels and 48 sub-slices each, recording a complete spectrum at every spatial pixel simultaneously — 90,000 spectra per exposure, each covering the full visible wavelength range from 480 to 930 nanometers. The result, for each galaxy, is a three-dimensional data cube: two spatial dimensions plus one spectral dimension, allowing the team to map exactly which physical process dominates at every location across each galaxy.

The nine Seyfert galaxies studied were Centaurus A, Circinus, IC 5063, NGC 1068, NGC 1365, NGC 1386, NGC 2992, NGC 4945, and NGC 5643. All are nearby active galaxies with well-documented AGN activity, making them ideal targets for a test of the new diagnostic method. The optical MUSE analysis was independently validated using archival X-ray data from NASA’s Chandra X-ray Observatory, providing a multi-wavelength cross-check on the team’s interpretation.

What the Data Revealed: Rings, Cones, and Perpendicular Shocks

Across all nine galaxies, the same geometric picture emerged. Extending outward from the galaxy plane, in two directions perpendicular to the galactic disc, are ionization bicones — cone-shaped volumes of gas lit up by the AGN’s hard radiation, a known feature of Seyfert galaxies. These were expected.

What was more surprising was the spatial relationship between the shocks and the star-forming regions. The shock-dominated zones consistently appeared perpendicular to the ionization bicones — not along the same axis as the AGN outflows, but in the plane of the galactic disc, where jets or winds pushing outward laterally compress the dense interstellar medium.

“The most interesting phenomenon about shocks is that they always go perpendicular to where the black hole’s injected outflows go,” Zhu said. “It is very common, and we see it consistently appearing across the whole nine galaxies.”

Beyond those shock-dominated central regions, at distances of approximately 0.8 to 6 kiloparsecs (about 2,600 to 19,600 light-years) from each galaxy’s center, the data showed rings or arcs of active star formation. The leading interpretation is causally direct: the AGN’s jets and winds compress the gas as they push into the interstellar medium; the compressed gas becomes dense enough to collapse under its own gravity; and that collapse produces new stars. The black hole is not quenching stellar birth — it is, in a geometrically organized way, triggering it. In galaxies with lower-power jets, the researchers note, AGN winds may also contribute alongside the jets.

What Astronomers Can Search for Next?

The results from nine nearby Seyfert galaxies raise an immediate and important question: how far does this pattern hold? The study’s scope is, by design, narrow — nine nearby galaxies at low redshift, all Seyfert class. The next scientific steps are to determine whether the same geometry appears in higher-luminosity AGN (quasars) and at higher redshifts, where both AGN activity and star formation rates were substantially elevated compared to the present-day universe. These questions will require instruments capable of resolving the same spatial detail at greater cosmic distances.

Future facilities are specifically suited to this work. The James Webb Space Telescope’s NIRSpec integral field unit can apply similar 3D diagnostic techniques in the infrared out to galaxies billions of light-years away. The forthcoming Giant Magellan Telescope, with its next-generation spectrograph GMACS (for which CfA is a contributing institution), will extend VLT-class spatial resolution to much fainter and more distant targets. If the Zhu et al. pattern holds at higher redshifts and higher AGN luminosities, it could reshape not just the interpretation of individual galaxy surveys but the entire AGN feedback prescription underlying galaxy formation simulations.

That prescription — which in simulations like IllustrisTNG and EAGLE is implemented as a single-mode, primarily thermal quenching mechanism — has been acknowledged by simulation teams themselves as a known simplification. The Zhu et al. data give that acknowledgment new observational urgency.

What Astronomers Say About the Finding: ‘Reshaping Their Surroundings’

The study carries the weight of both halves of the CfA’s institutional expertise: Zhu’s observational and diagnostic work, grounded in MUSE data and the new 3D emission-line method, combined with Sutherland’s theoretical shock and photoionization models, which allow the team to cross-validate what the observations show against what the physics predicts. Kewley — whose prior work establishing the Kewley et al. (2001) theoretical starburst demarcation line on the BPT diagram is itself foundational to the field — described the finding as revealing a previously underappreciated dynamic: the black hole simultaneously consuming material falling inward and reorganizing the galaxy around it through outflows.

The full author list also includes Dominika Krol, Giuseppina Fabbiano, Lars Hernquist, Anna Trindade Falcão, Martin Elvis, and Riccardo Middei, reflecting the multi-institutional scope of the project.

The finding does not mean negative AGN feedback is wrong. The two processes — quenching and triggering — are not mutually exclusive; they may occur simultaneously in different spatial regions of the same galaxy, as some prior studies of individual systems have suggested. What the Zhu et al. survey adds is statistical grounding: this is not a one-off anomaly but a consistent geometric pattern across nine independently studied active galaxies. That consistency is the study’s primary contribution — and the strongest reason why simulation teams will need to take it seriously.

Frequently Asked Questions

Can a black hole really trigger star formation? Isn’t that the opposite of what they do?

Yes — and the counterintuitive nature of the finding is precisely why it matters. The dominant model of AGN feedback predicts that an active black hole will heat and expel a galaxy’s star-forming gas, shutting down stellar birth. This study found the opposite: in all nine galaxies surveyed, the black hole’s jets and winds appear to compress gas in the galactic plane, triggering star formation in rings at distances of roughly 2,600 to 19,600 light-years from the center. The two processes — suppression and triggering — are not mutually exclusive and may operate simultaneously in different regions of the same galaxy.

What is AGN feedback and why does it matter for understanding galaxies?

AGN stands for active galactic nucleus — the bright central region of a galaxy powered by a supermassive black hole that is actively pulling in surrounding gas. “Feedback” refers to the energy that process releases back into the galaxy: radiation, winds, and jets that can heat, compress, or expel the gas from which new stars would otherwise form. This feedback is one of the primary mechanisms astronomers use to explain why the universe’s most massive galaxies stopped forming stars while lower-mass galaxies continued. Every major cosmological simulation — including IllustrisTNG and EAGLE — currently models AGN feedback primarily as a quenching (“negative”) process. This new study challenges whether that is the complete picture.

Why couldn’t astronomers see this positive feedback effect before?

The key obstacle was a technical one: the standard diagnostic tool used to classify emission-line regions inside galaxies — the BPT diagram, developed in 1981 — cannot reliably separate the signatures of AGN photoionization from those of fast shocks. Because shock signatures look like AGN signatures on the BPT diagram’s 2D plot, prior studies may have misclassified shock-dominated regions as purely AGN-dominated, masking the compression that leads to star formation. The new 3D diagnostic developed by Zhu and colleagues adds a third emission-line ratio axis that specifically separates shocks from AGN photoionization, making the star-forming rings visible for the first time at this level of detail across a multi-galaxy sample.

What does this mean for galaxy formation models and simulations?

It means the AGN feedback prescription in major cosmological simulations — which implement AGN primarily as a star-formation suppressor — may be systematically underpredicting how much star formation active black holes actually enable. IllustrisTNG, EAGLE, and SIMBA all rely on this negative-feedback-dominant prescription, and simulation teams have already begun acknowledging that their subgrid models of AGN physics need updating. The Zhu et al. survey’s statistical consistency across nine different galaxy systems gives those calls for revision new observational weight. Follow-up work with JWST and the Giant Magellan Telescope will determine whether the same pattern holds at higher redshifts, where AGN and star formation activity was substantially more intense.

“Common Excitation Patterns of Star Formation, Active Galactic Nuclei, and Shocks in Seyfert Galaxies” was published September 14, 2026, in The Astrophysical Journal (DOI: 10.3847/1538-4357/ae9956; arXiv:2609.15913). The study was led by Peixin Zhu in collaboration with Lisa Kewley, Dominika Krol, Giuseppina Fabbiano, Lars Hernquist, Ralph Sutherland, Anna Trindade Falcão, Martin Elvis, and Riccardo Middei. Observations were made using VLT/MUSE at the European Southern Observatory’s Paranal Observatory, Chile, with supporting data from NASA’s Chandra X-ray Observatory.