How Kerr superradiance allows rotating black holes to build macroscopic quantum clouds, lose angular momentum to fields almost too light to detect on Earth, and potentially broadcast their quantum spectra through gravitational waves.
A black hole is usually imagined as the ultimate absorber. Matter can fall through its horizon, light can disappear into it, and no classical signal can return once it crosses the boundary. A rotating black hole complicates that picture. Kerr spacetime contains a reservoir of rotational energy outside the horizon, and under the right conditions an incident wave can leave the black hole with more energy than it had before the encounter. If the field also has a small mass, gravity can trap the amplified wave near the black hole and send it through the amplification process again. What begins as a tiny fluctuation can then grow exponentially into a macroscopic bosonic cloud surrounding the horizon. This phenomenon, known as the black-hole superradiant instability, transforms an astrophysical black hole into something unexpectedly similar to an atom. [1][2][3]
The analogy is not superficial. The black hole acts as a gravitational nucleus, while a massive bosonic field occupies discrete bound states whose weak-coupling spectrum resembles the spectrum of hydrogen. Unlike an ordinary atom, however, the “electron cloud” can contain an enormous occupation number of bosons coherently populating the same quantum state, and the nucleus is a rotating event horizon capable of transferring its angular momentum into the cloud. The result is often called a gravitational atom. Recent work has pushed this idea in several directions at once: a 2025 study canonically quantized the unstable scalar field around Kerr, 2026 work has refined the gravitational-wave spectrum of axion clouds, another 2026 proposal suggests that these systems could behave as gravitational-wave amplifiers through stimulated emission, and an August 2026 preprint has investigated when self-interacting clouds can be treated as Bose-Einstein condensates in the full Kerr geometry. [1][10][11][12]
The mechanism begins with the geometry of Kerr spacetime. Throughout this article, use natural units with $G=c=\hbar=1$. A Kerr black hole is specified by its mass $M$ and angular momentum $J$, with the usual rotation parameter
$$ a=\frac{J}{M} $$
The outer event horizon is located at
$$ r_+=M+\sqrt{M^2-a^2} $$
and rotates with angular velocity
$$ \Omega_H=\frac{a}{2Mr_+} $$
For $|a|<M$, the horizon is regular; the extremal limit is approached as $|a|\rightarrow M$. Outside the event horizon lies the ergoregion, where frame dragging becomes so strong that an observer cannot remain stationary relative to infinity. The time-translation Killing vector that defines conserved energy at infinity becomes spacelike there, allowing field configurations with negative Killing energy relative to an observer at infinity. That unusual energy bookkeeping is what makes rotational-energy extraction possible. The modern quantum treatment of a scalar field on Kerr uses precisely this horizon angular velocity through the combination $\omega-m\Omega_H$. [1]
This is the wave analogue of the Penrose process. In the particle version, a process inside the ergoregion can send negative Killing energy into the horizon while another fragment escapes carrying more energy than the original object. Superradiance implements the same basic logic with fields. Part of the wave effectively transports negative Killing energy through the horizon, and energy conservation then requires the outgoing component to carry away more positive energy than the incoming wave possessed. The black hole pays the difference from its rotational energy. [3]
To make the argument quantitative, consider a massive scalar field $\Phi$ of mass $\mu$. On the Kerr background it obeys the curved-spacetime Klein-Gordon equation
$$ (\Box-\mu^2)\Phi=0 $$
The stationarity and axial symmetry of Kerr allow the field to be separated into modes of the form
$$ \Phi=e^{-i\omega t}e^{im\phi}S_{\ell m}(\theta)R_{\ell m}(r) $$
where $\omega$ is the mode frequency, $m$ measures its azimuthal angular momentum, $\ell$ labels the angular structure, $S_{\ell m}$ is a spheroidal harmonic, and $R_{\ell m}$ contains the radial dynamics. The horizon does not respond simply to $\omega$. Because the horizon itself rotates, the physically relevant near-horizon frequency is shifted to
$$ k_H=\omega-m\Omega_H $$
as appears directly in the asymptotic Kerr mode solutions. [1]
The energy flux through the horizon has a sign controlled by this shifted frequency. Suppressing positive normalization factors, its structure can be represented schematically as
$$ \frac{dE_H}{dt}\propto\omega(\omega-m\Omega_H)|A_H|^2 $$
For an ordinary positive-frequency mode with $\omega>m\Omega_H$, the horizon absorbs positive energy. When instead
$$ 0<\omega<m\Omega_H $$
the horizon flux becomes negative. The black hole loses energy and angular momentum while the exterior field gains them. This is the superradiant condition. It immediately explains why rotation is essential and why modes corotating with the hole, with positive $m$, are the ones capable of extracting its spin. [1][2]
The process does not violate conservation of energy. The amplified wave is powered by the black hole's rotational reservoir. In fact, the extraction eventually weakens the very condition that made amplification possible because $\Omega_H$ decreases as the black hole spins down.
Superradiant scattering by itself does not yet produce a gravitational atom. An amplified massless wave can simply escape to infinity. A massive field changes the story because sufficiently low-frequency modes cannot freely propagate arbitrarily far from the black hole. For a bound state, the real part of the field frequency satisfies approximately
$$ \omega_R<\mu $$
so the field decays exponentially at large radius instead of behaving as a freely propagating wave. Gravity attracts the field toward the black hole while the mass term provides an effective outer confinement. The result is a quasibound state trapped outside the horizon. Dolan's numerical analysis of the massive Klein-Gordon equation demonstrated that these bound states become unstable when they also satisfy the superradiant condition. [2]
The combination is the essential feedback loop. The Kerr horizon amplifies the field, the massive-field potential prevents the amplified mode from escaping, and the bound field returns to interact with the rotating black hole. The occupation of the unstable mode grows repeatedly, creating the natural analogue of the “black-hole bomb” proposed in early studies of superradiance, except the boson's own mass replaces an artificial reflecting mirror. [2][3]
The atomic analogy becomes precise when the boson's Compton wavelength is large compared with microscopic scales but comparable to the gravitational size of the black hole. Define the gravitational fine-structure constant
$$ \alpha=M\mu $$
in natural units. Equivalently, $\alpha$ measures the ratio between the black hole's gravitational radius and the boson's Compton wavelength. When $\alpha\ll1$, the bound states are nonrelativistic over much of the cloud and the spectrum becomes approximately hydrogenic. To leading order,
$$ \omega_n\simeq\mu\left(1-\frac{\alpha^2}{2n^2}\right) $$
where $n$ is the principal quantum number. The cloud radius scales parametrically as
$$ r_{\rm cloud}\sim\frac{n^2M}{\alpha^2} $$
so weakly bound clouds can extend many gravitational radii beyond the event horizon. This hydrogen-like structure has been developed into a detailed spectroscopy of gravitational atoms, including relativistic, fine-structure, and hyperfine corrections generated by the black hole's rotation. [4][5]
The resemblance to hydrogen has a crucial difference. In an ordinary atom, the fine-structure constant describes electromagnetic binding between an electron and a nucleus. Here the interaction is gravitational, the central object is spacetime itself, and the orbiting object is a quantum field. For ultralight bosons, $\alpha$ can become order $0.1$ around astrophysical black holes even though gravity is microscopically weak, simply because both the black hole mass and the boson's wavelength are enormous by particle-physics standards. Superradiance is most effective when these scales become comparable. [3][13]
The bound-state frequency is not perfectly real. Write it as
$$ \omega=\omega_R+i\Gamma $$
so the mode evolves in time as
$$ \Phi\propto e^{\Gamma t} $$
when $\Gamma>0$. A positive $\Gamma$ therefore signals a superradiant instability. In the weak-coupling regime, scalar growth rates are extraordinarily sensitive to $\alpha$. Up to mode- and spin-dependent coefficients, the dimensionless rate follows the steep scaling
$$ M\Gamma\propto\alpha^{4\ell+5} $$
for the relevant hydrogenic regime. This dependence explains why superradiance selects narrow combinations of black-hole mass and boson mass rather than acting equally strongly on every ultralight particle. The lowest corotating scalar modes are generally the fastest-growing; Dolan found the $\ell=m=1$ state to be the dominant scalar instability over the important parameter range he studied. [2][12]
As the cloud grows, it removes angular momentum from the black hole. A field quantum in a mode $(\omega,m)$ carries energy $\omega$ and axial angular momentum $m$ in natural units, so the ratio is
$$ \frac{dJ}{dE}=\frac{m}{\omega} $$
The superradiant process therefore drives the Kerr black hole downward in spin until the horizon slows enough that the dominant cloud level reaches approximately
$$ \omega_R=m\Omega_H $$
At this saturation condition, the superradiant gain disappears. The gravitational atom is self-limiting: the field grows by extracting precisely the resource, black-hole rotation, that allows the instability to operate. Current superradiance searches exploit this predicted spin-down by looking for regions of the black-hole mass-spin plane that should be depleted if a boson with a particular mass exists. [6][13]
This is where the gravitational atom becomes a particle detector. Laboratory experiments struggle to probe particles whose masses are fantastically small and whose couplings to ordinary matter may be almost nonexistent. A Kerr black hole does not need the particle to interact electromagnetically. Gravity alone can provide the bound state, while rotation supplies the amplification. If the boson's Compton wavelength matches the appropriate black-hole scale, superradiance can magnify an initially negligible field into an astrophysically relevant cloud. [3][6]
The cleanest indirect signal would be the absence of highly spinning black holes in parts of the mass-spin plane where a particular ultralight boson should have efficiently extracted their angular momentum. Such constraints require care because black-hole ages, natal spins, accretion histories, binary interactions, and boson self-interactions affect the prediction. Earlier analyses using gravitational-wave and X-ray spin measurements already constrained portions of ultralight-boson parameter space under specified astrophysical assumptions. [6][8]
The observational story became substantially more current in August 2026. A new analysis using 257 binary-black-hole mergers in the public GWTC-5 catalog reported no evidence for a superradiant axion signal and, within its population model and superradiance assumptions, excluded axion masses from approximately $1.7\times10^{-14}$ eV to $3.3\times10^{-12}$ eV at 95% confidence. The authors explicitly modeled the natal-spin population and marginalized over merger delay times rather than assuming that every black hole had the same evolutionary history. This is a very recent preprint, so its precise bounds should be interpreted as a current research result rather than a final universal exclusion, especially because strong self-interactions or different astrophysical systematics can modify superradiant evolution. [13]
The field can also reveal itself directly through gravitational radiation. A boson cloud is not simply a static spherical halo. Its occupied levels carry angular structure, and time-dependent multipole moments can source gravitational waves. Two broad mechanisms are especially important. Bosons in the cloud can annihilate into gravitational radiation, producing a long-lived signal related to roughly twice the boson oscillation frequency, while transitions between different gravitational-atom levels can radiate at the energy difference between the levels. [4][7][10]
For a transition between levels $i$ and $f$, the characteristic gravitational-wave frequency is controlled by
$$ \omega_{\rm GW}\simeq\omega_i-\omega_f $$
whereas pair-annihilation channels occur at frequencies controlled approximately by the sum of the participating boson energies. Because gravitational-atom levels can be extremely narrow and the clouds can live for long times after saturation, the resulting signals can be nearly monochromatic rather than short chirps like ordinary compact-binary mergers. LIGO-Virgo-KAGRA has already performed searches specifically targeting continuous or long-lived signals from boson clouds. An O3 all-sky scalar-cloud search found no evidence for such signals, and newer O4 analyses have begun directed searches for vector-boson clouds around known black holes and merger remnants. [7][14]
Binary companions make the spectroscopy even richer. Their tidal field can perturb the cloud and drive resonant transitions between gravitational-atom levels. A 2025 study calculated gravitational waves generated during such level transitions and found that stellar-mass gravitational atoms with scalar masses around $10^{-12}$ eV can produce signals near the decihertz band in particular transition scenarios, a regime relevant to proposed future detectors such as BBO and DECIGO. [9]
The spectroscopy is becoming precise enough that researchers now worry about effects that would be tiny in a first approximation. An August 2026 analysis developed relativistic perturbation theory for frequency shifts in gravitational waves from self-interacting axion clouds. In these systems, the emitted frequency is affected not only by the hydrogenic binding energy but also by relativistic corrections, cloud self-gravity, nonlinear axion interactions, and the occupation of multiple modes. These corrections matter because continuous-wave searches depend sensitively on predicting how the signal frequency evolves over long observation times. [10]
This is the astrophysical equivalent of moving from recognizing the gross spectrum of hydrogen to resolving its fine structure. If gravitational atoms exist, simply detecting an excess of continuous gravitational radiation would be only the beginning. The detailed frequency pattern could encode the boson mass, the Kerr spin, the quantum numbers of occupied states, self-interactions within the cloud, and perturbations from nearby objects. The black hole would become not merely a detector of new fields, but a spectroscopic instrument for them. [5][9][10]
There is also a genuinely quantum question hiding beneath the familiar classical description. Most calculations of superradiant cloud growth treat the scalar field as a classical wave. That approximation is extremely effective once the occupation number is enormous, but it leaves a conceptual gap at the beginning of the instability. If there is initially no classical cloud, what exactly is growing? How should one describe particle production, the unstable discrete modes, and the transfer of energy and angular momentum using quantum field theory rather than classical amplitudes?
A 2025 preprint by Fu, Omiya, Tanaka, Tong, Wang, and Zhu addressed this problem by performing a canonical quantization of a massive real scalar field around Kerr. Their construction includes both continuous-frequency scattering modes and discrete superradiant modes. The unstable discrete sector is unusual because its complex frequencies do not behave like the standard stable harmonic oscillators encountered when quantizing a field in flat spacetime. The authors nevertheless construct a quantum framework in which particle-number growth and the energy and angular momentum of the field can be followed consistently. [1]
For the unstable sector, the particle-number evolution contains terms with the characteristic behavior
$$ \langle N(t)\rangle\propto\sinh^2(\Gamma t) $$
so at late times the occupation grows exponentially. Importantly, the authors argue that the cloud growth occurs independently of the choice of initial state: superradiance is not merely the classical amplification of an already populated cloud. The quantum instability itself supplies a route from microscopic fluctuations to macroscopic occupation. They also place Hawking radiation, superradiance, adiabatic spin-down, and some effects of self-interaction within a common QFT-in-curved-spacetime framework. [1]
This quantum description also exposes why Kerr is more subtle than simply putting an ordinary quantum field next to a gravitational potential. The conserved energy associated with time translations is not positive definite throughout the ergoregion, and superradiant modes are precisely those for which the horizon-shifted frequency
$$ k_H=\omega-m\Omega_H $$
is negative. In the 2025 quantization, the authors emphasize that the unstable discrete modes require an unconventional organization of creation and annihilation operators and that the Hamiltonian does not possess the same straightforward lowest-energy vacuum structure one expects for a stable field in Minkowski space. [1]
This is a concrete example of a broader lesson from quantum field theory in curved spacetime: the concept of a “particle” depends on how field modes are divided into positive- and negative-frequency sectors, and spacetime geometry can make that division nontrivial. Kerr adds rotation, a horizon, an ergoregion, and unstable bound modes to the problem simultaneously. The gravitational atom therefore lies at an unusual intersection where particle physics, black-hole thermodynamics, classical instability, and QFT cannot be cleanly separated from one another. [1]
The simplest gravitational-atom picture assumes that the bosons interact only gravitationally. Axions and many axion-like particles can instead possess self-interactions. Even a weak nonlinear interaction can become significant once an enormous number of particles occupy the cloud, because the field amplitude becomes macroscopically large. This changes the spectrum, growth rate, saturation mechanism, and gravitational-wave emission. A 2025 full-Kerr study of self-interacting scalar clouds found that self-interaction can suppress cloud growth and shift both cloud oscillation frequencies and the gravitational-wave frequencies produced by the system, potentially weakening spin-based particle constraints that assume a free scalar field. [8]
An August 2026 preprint approached the same problem from the language of Bose-Einstein condensation. Starting with a scalar field containing quartic self-interaction in Kerr, it derived an effective Gross-Pitaevskii description and reported that the resulting condensate is toroidal rather than spherically shell-like, with its density vanishing on the rotation axis for the modes studied. The work also found that rotation geometrically modifies the effective self-interaction near the horizon. Because this result is a very recent theoretical preprint, its detailed conclusions should be viewed as part of an active research direction rather than settled phenomenology, but it makes the term “gravitational atom” even more literal: the macroscopic cloud can be studied using ideas familiar from quantum many-body condensates while remaining embedded in strongly curved Kerr spacetime. [12]
An even more speculative 2026 proposal suggests that gravitational atoms may do something ordinary atoms are famous for: stimulated emission. In an ordinary laser or maser, incoming radiation resonant with a transition can stimulate an excited system to emit additional radiation coherently. A June 2026 preprint formulated an analogous mechanism for boson clouds, proposing that an ambient gravitational-wave field resonant with transitions between cloud levels can stimulate enhanced gravitational-wave emission. The authors derive selection rules and threshold conditions and describe the system as a possible natural gravitational-wave amplifier. [11]
The proposed mechanism is theoretically intriguing but has not been observed. Its significance is the way it extends the atomic analogy. A Kerr black hole can provide the central gravitational potential, superradiance can populate excited bosonic levels, and a passing gravitational wave can in principle interact with those levels. If stimulated transitions can become astrophysically important in realistic environments, a gravitational atom would behave less like a passive cloud and more like an active quantum-optical system built from spacetime, a black hole, and an ultralight field. [11]
The entire phenomenon depends on a remarkable hierarchy of scales. The event horizon may span kilometers or millions of kilometers, the bosonic wave function may extend far beyond it, and the particle itself may have a mass many orders of magnitude below anything that conventional accelerators could efficiently probe. Yet the underlying dynamics are governed by the same quantum principles that produce discrete atomic levels: wave equations, boundary conditions, angular momentum quantum numbers, occupation numbers, and transitions between eigenstates.
Kerr geometry provides the part ordinary atoms do not possess. Frame dragging shifts the field frequency at the horizon, the ergoregion permits negative Killing-energy flux, and the horizon absorbs precisely the component required for rotational-energy extraction. A massive quantum field then converts that one-pass amplification into an instability by trapping the mode. The bosonic nature of the field allows a single level to acquire enormous occupation, and backreaction eventually spins the black hole down until the resonance shuts itself off. [1][2][3]
The result is a system that is simultaneously a solution of general relativity, a bound-state problem in quantum field theory, a possible dark-matter detector, and a source of gravitational-wave spectroscopy. It is difficult to find another physical system in which the rotation of an object millions of times larger than an atom can selectively amplify a particle field according to the integers $\ell$ and $m$, populate hydrogen-like quantum levels, and potentially encode those levels into radiation detectable across astronomical distances.
The phrase black-hole atom therefore describes more than a useful analogy. Kerr superradiance creates a regime in which the quantum mechanics of bound states is written directly into rotating spacetime. If ultralight bosons exist, some of the universe's black holes may already be surrounded by enormous quantum clouds, gradually surrendering their spin while those clouds wait to reveal themselves through missing black-hole spins, narrow gravitational-wave lines, level transitions, or more exotic collective effects that current theory is only beginning to understand.