Deep divefrom Episode 4New physics & chemistry30 September 202621 min

The Black-Hole Bomb That Didn't Go Off

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: End state of the experimental black-hole bomb (arXiv)

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1You heard the trailer — this is the whole story. Welcome to the Hastrup Review of Science — where artificial intelligence reads real science, so you don't have to pretend you understood the paper. One paper today: a black-hole bomb built in a tank of superfluid helium, and what happens when it is allowed to run to the end.

Host 2The paper is End state of the experimental black-hole bomb, by Patrik Švančara and twelve colleagues, led from the University of Nottingham with collaborators in Manchester, London, Newcastle, Grenoble, Brazil and Canada. It is a preprint on the arXiv, not yet peer reviewed. The group, around Silke Weinfurtner, has spent years building black-hole analogues out of flowing fluids.

Host 1And nothing actually explodes.

Host 2Nothing explodes, and there is no black hole. There is a whirlpool in liquid helium below two kelvin, and a wave sloshing around it that grows until something stops it. The paper is about what that something is.

Energy from a spinning black hole

Host 1Start with the name. Why would anyone call something a black-hole bomb?

Host 2As established background, it goes back to Roger Penrose, who noticed around nineteen sixty-nine that a rotating black hole stores energy that can, in principle, be taken out. Around a spinning black hole there is a region called the ergoregion, where spacetime is dragged around so strongly that nothing can stay still. Inside it, things can carry negative energy as measured from far away.

Host 1Negative energy sounds like a trick.

Host 2It is bookkeeping, but real bookkeeping. In the Penrose process, as the paper describes it, an object falls in and splits in two: one fragment is captured by the black hole, and the other escapes with more energy than the whole object had. The black hole pays for the difference by spinning down slightly.

Host 1And that works for waves too?

Host 2Yes, and that is superradiance. Yakov Zel'dovich showed in the early nineteen seventies that a rotating body can amplify waves scattering off it, and a spinning black hole does the same. A wave can come back stronger than it went in, with the extra energy taken from the spin.

Host 1So where does the bomb come in?

Host 2In nineteen seventy-two, William Press and Saul Teukolsky asked what happens if you put a mirror around the black hole. Each time the wave bounces back in, it is amplified again, so its amplitude grows exponentially. They called it the black-hole bomb, and the name stuck.

Why astrophysicists care

Host 1Nobody is building mirrors around black holes. Why does this matter outside a thought experiment?

Host 2Because nature can supply the mirror. If a field has mass, like a hypothetical ultralight particle such as an axion, that mass can trap it near the black hole, and the paper also lists plasma and other environments as partial confining mechanisms. The trapped field then grows into a long-lived cloud, extracting energy and angular momentum from the black hole.

Host 1And that would be observable?

Host 2That is the hope. Such a cloud should produce distinctive gravitational waves, which is why the paper calls rotating black holes powerful probes of physics beyond the Standard Model. Current and future detectors are already searching for these signals.

Host 1Which only works if you know how the growth ends.

Host 2Exactly. Linear theory says exponential growth forever, but real growth has to stop, and how it stops changes what you would observe. The paper cites theory where particle production saturates the instability early, and other work where attractive self-interactions make the cloud collapse back into the black hole, an event nicknamed a bosenova.

Black holes in the lab

Host 1You cannot run that experiment on a real black hole. So what do physicists do?

Host 2They use analogue gravity. Waves travelling on a flowing medium behave as if they were moving through a curved spacetime, and the flow can mimic horizons and ergoregions. It says nothing about gravity itself, but it tests the wave physics, which is the part that matters for superradiance.

Host 1Had superradiance been seen that way?

Host 2Yes, on several platforms. The Nottingham group saw it in surface waves around a draining water vortex in twenty seventeen, others saw it with light in nonlinear media, and a team led from Glasgow and Southampton saw electromagnetic waves amplified by a spinning rotor in an electric circuit.

Host 1And the bomb itself?

Host 2That circuit team, Cromb and colleagues, added a resonator in twenty twenty-five to keep reflecting the amplified wave, and reported the first experimental black-hole bomb. But their runaway ended because the rotor, part of a modified electric motor, lost energy until it dropped below the threshold for amplification. The system ran out of fuel, so it never reached a nonlinear end state.

Host 1Which is the open question.

Host 2Yes: if the spinning object keeps its energy, what stops the runaway? The paper calls that nonlinear evolution poorly understood, despite its central role in models of boson clouds around black holes.

The set-up

Host 1So let us build it. What is in the tank?

Host 2Helium four, held at one point seven zero kelvin, where it is a superfluid. A cylindrical container seventy-four point six millimetres across sits partly submerged in a helium bath, filled to a depth of twenty-one point six millimetres, with a ten-millimetre drain in the middle of the floor. An eight-blade propeller under the drain, turning at two hundred and forty revolutions per minute, pulls helium down through the drain and refills the zone near the walls.

Host 1So it is a bathtub drain that never empties.

Host 2Right, a recirculating bathtub vortex, with a thin hollow funnel running from the surface down to the drain. A custom flow conditioner suppresses rotation of the fluid as a solid body, so the flow swirls almost like an ideal vortex, faster near the centre. The solid-body part contributes less than six percent of the swirl in the region they study.

Host 1Why superfluid helium and not water?

Host 2Its kinematic viscosity at that temperature is tiny, eight point nine times ten to the minus nine square metres per second. That means a wave with a millimetre-scale wavelength takes on the order of a hundred seconds to die away, comparable to the whole experiment. The fluid's properties also let them build a vortex strong enough to sustain repeated amplification.

Host 1How strong?

Host 2They fit the circulation at nineteen point two square centimetres per second. In a superfluid, circulation comes in fixed quanta, each carried by a thin quantum vortex line, and that value corresponds to about one hundred and twenty-one thousand quanta. So the central funnel behaves as a giant vortex built from over a hundred thousand tiny ones.

Where the analogy comes in

Host 1Where is the black hole in all this?

Host 2For long surface waves, the swirling flow acts like a rotating spacetime with two space dimensions and one time dimension. Near the vortex the flow is fast enough to drag waves around with it, which plays the role of the ergoregion. The container wall, at thirty-seven point three millimetres radius, is the mirror.

Host 1Walk me through the amplification.

Host 2They use the WKB approximation, which treats the waves locally as plane waves with a slowly changing wavelength. It reveals a band of radius and frequency where waves cannot propagate, only decay, and that band acts as a potential barrier between the vortex and the outer region. Part of a wave tunnels through, the part near the core is dragged into co-rotation and carries negative energy, and when the vortex absorbs it, the wave outside comes back amplified.

Host 1Then it bounces off the wall and does it again.

Host 2Yes. The reflection coefficient at the barrier is bigger than one, and the growth rate is the logarithm of that coefficient divided by the round-trip time between barrier and wall. The most efficient amplification is for resonant bound states, waves trapped between the barrier and the wall.

Host 1Can they be sure it is superradiance and not something else in the vortex core?

Host 2Not fully. The core contains many quantum vortices, and oscillations of that vorticity can also drive an instability; modelling that in a three-dimensional, dispersive system is beyond this paper. In a simpler shallow-water model, their earlier work found the ergoregion dominates when the core is hollow, as it is here, but that is an argument from a related model, not a measurement.

Seeing the waves

Host 1How do you even measure a wave on helium inside a cryostat?

Host 2They film the surface from above through a flat window, against a printed sinusoidal pattern, and the waves distort the pattern the way a lens would. The technique, synthetic Schlieren imaging, reconstructs the surface height point by point with a sensitivity of about half a micrometre. The camera records twenty-one thousand four hundred frames at two hundred frames per second, one thousand five hundred and thirty-six pixels square.

Host 1Then what?

Host 2They resample onto a polar grid of two hundred and fifty-six radial by one hundred and twenty-eight angular points, and Fourier transform in time and angle. Each wave then gets two labels: its frequency, and its azimuthal number, how many times the pattern repeats around the vortex. The field of view is a ring from seven to thirty-one point five millimetres in radius, missing both the drain region and the wall.

Host 1And the background flow, did they measure that separately?

Host 2They read it off the waves themselves. Before the instability takes over, the propeller's mechanical noise excites small waves, and the shape of that noise spectrum reveals the flow, a method from their twenty twenty-four Nature paper on this vortex. The radial inflow at the surface turns out small; the swirl dominates.

The bomb goes off

Host 1So they switch on the vortex. What happens?

Host 2A rotary sloshing wave appears, precessing around the vortex core, visible by eye through the glass. It has azimuthal number one, a single bulge going around, and oscillates at three point four seven hertz. Its amplitude grows exponentially at zero point zero eight one per second, give or take zero point zero zero one.

Host 1Put that growth rate in human terms.

Host 2It means the amplitude doubles roughly every eight and a half seconds. Keep that up for over a minute and you go from micrometre-scale noise to a wave you can see.

Host 1And then?

Host 2At about eighty seconds after the vortex is switched on, the growth stops. The sloshing amplitude, averaged over radius, saturates at three point zero six millimetres, plus or minus zero point zero five, on a layer of helium about twenty-two millimetres deep. The system then sits in what they call a long-lived non-equilibrium steady state: energy is still available from the vortex, but the wave no longer grows.

Host 1Is that the only wave?

Host 2No. Waves at two, three, up to eight times the sloshing frequency, with matching azimuthal numbers, also grow and saturate. The second harmonic grows about twice as fast and higher ones faster still, which is what you expect if the primary wave is driving them.

Host 1Does theory get the sloshing frequency right?

Host 2The WKB model predicts three point nine seven hertz for the lowest bound state, reasonable given the approximations, but the authors expect WKB to break down for a wave as large as the container. It also shows that the frequency is set by the flow near the outer edge, while the growth rate depends on the flow near the drain, which the camera cannot see. So they built a second model for the edge.

A word from our sponsor

Host 1Do you have problems keeping things cold when you cannot afford a helium cryostat? Then you are in luck.

Host 1The Hastrup Review of Science is brought to you by ice cube trays. Fill them with water, leave them in the freezer overnight, and in the morning you have something firm, cold and reliably shaped.

Host 2I keep a couple in the lab freezer. When things get too hot between runs, I just pop one out and slip it in where it is needed.

Host 1They come in silicone or plastic, in cubes, spheres and novelty shapes, and they are very reasonably priced. They are widely available.

Host 2And unlike superfluid helium, what you put in them stays put. Nothing creeps up the sides and nothing drains away while you are not looking.

Host 1Ice cube trays. Our referral link is in the show notes.

A model for the edge

Host 1Back to the paper. What is the second model?

Host 2Near the wall the swirl is slow, about a tenth of the shallow-water wave speed, so they treat both the vortex and the wave's own nonlinearity as small corrections to still fluid, with one small parameter of about zero point one. They write down the physics of surface waves up to fourth order in wave amplitude, which keeps three- and four-wave interactions. The modes of still fluid in a cylinder are known exactly, and the vortex and the nonlinearity shift them order by order.

Host 1What does that buy them?

Host 2First, the frequency: three point four four hertz, against three point four seven measured. Second, a prediction that the frequency should drift as the wave grows, with the shift proportional to the amplitude squared. For the size of that shift they use a nineteen eighty-four calculation by John Miles for waves in a cylinder, which applies at their order of approximation.

Host 1And did the frequency drift?

Host 2It did. Following the sloshing mode's spectrogram through time, they see a cumulative drift of zero point zero nine eight hertz, which they call fully compatible with the model, though the predicted and observed frequencies sit at a small offset. It is a direct sign that the wave has left the linear regime.

Host 1What else does the model predict?

Host 2Shapes. The big wave should push the average surface into a static dip, which analogue gravity people call backreaction, and it should drive a second harmonic with two bulges at twice the frequency. At sixty-five seconds, mid-growth, the measured radial profiles of the dip and the primary wave agree satisfactorily with the model.

Host 1And the second harmonic?

Host 2Less good. Much of that wave sits near the vortex core, where treating the vortex as a small correction fails; its shape fluctuates in time; and the model allows only one higher mode when the data show several. It gets roughly the right amplitude near the edge, where the approximation should work best.

Host 1Does the model explain why the growth stops?

Host 2No, and it is worth being clear about that. The growth rate is put into the model by hand, as a phenomenological parameter, because the approximation neglects what happens inside the core. The model explains the frequency drift and the mode shapes; it does not derive the saturation.

Wave mixing

Host 1So what does stop it?

Host 2Their candidate is wave mixing. Surface waves interact nonlinearly: two waves can combine into a third, three into a fourth, and so on, as long as the frequencies, and so the energies, add up. If the sloshing wave keeps feeding energy into other waves, those waves act as a drain on it.

Host 1How do you show waves are interacting rather than just coexisting?

Host 2With correlations. They multiply the amplitudes of the chosen waves, average over space, and subtract everything that could come from lower-order correlations, leaving what is called the connected part, or Ursell function. That isolates processes where all the waves genuinely take part together, and they normalise it by all correlations of that order.

Host 1And the result?

Host 2For three-wave mixing, two sloshing waves making the second harmonic, and four-wave mixing, three making the third, the normalised connected correlation climbs and saturates near one once the steady state sets in. Early on it is noisy because the waves are small, but late in the run these interactions are clearly resolved.

Host 1Is it only the harmonics?

Host 2No, and this is the more interesting map. When they let one incoming frequency and the outgoing frequency vary freely, the strong correlations fall along lines where frequency is conserved, extending well beyond simple multiples of the sloshing frequency and across all azimuthal numbers. Five- and six-wave processes show the same pattern, but weaker and embedded in noise.

Host 1So energy spreads out into a whole spectrum.

Host 2That is the picture: the end state is shaped by the unstable wave's self-interaction plus many energy-conserving mixing processes that spread its energy into secondary waves. But the authors state plainly that this energy drain stabilising the steady state is a hypothesis. Examining the mechanism in detail, they say, is beyond this work.

Why it matters

Host 1Step back. Why does this matter?

Host 2The principle is that a superradiant runaway does not have to continue until the rotating object is exhausted. Here the vortex keeps running, and the evidence points to nonlinear wave interactions capping the growth and holding the system in a steady state. Unlike the earlier circuit experiment, which ran out of fuel, this black-hole bomb reaches that kind of end state.

Host 1What changes, and for whom?

Host 2For theorists modelling boson clouds around black holes, it is a controlled example where the end state is decided by the field's self-interactions, which is exactly the uncertain ingredient in their predictions. The authors note their result resembles the saturation scenario rather than the bosenova collapse, while a cold-atom analogue instead ends with angular momentum piling up at the drain. For experimentalists, it is a platform where the nonlinear phase can be measured wave by wave.

Host 1And what does it not mean?

Host 2It does not tell us how clouds around real black holes end, or whether axions exist at all. A surface wave on helium has its own nonlinearity, set by fluid mechanics, not by any particle physics, so the analogy is in the structure of the problem, not in the numbers. At most, it shows which kinds of nonlinear behaviour are possible and how to measure them.

Caveats

Host 1What are the honest caveats?

Host 2First, the data. The results shown come from one recording of about a hundred seconds, and the authors say that, given the limited dataset, a detailed assessment of statistical significance could not be achieved; the correlations converge within five percent and bootstrap errors are at that level, which is internal consistency, not repeated runs.

Host 1Second?

Host 2The saturation mechanism is a hypothesis: the model fits frequencies and shapes, but the growth rate goes in by hand. And the region where amplification actually happens, near the core, is outside the field of view, so the reflection coefficient is never measured, and a contribution from core vorticity cannot be fully excluded.

Host 1Anything in the numbers themselves?

Host 2A small inconsistency: the main text gives the measured frequency as three point four seven hertz, while the supplement calls three point four four the measured value and quotes a seventeen percent gap to the WKB prediction, larger than those numbers imply. It does not change the conclusions, but it is the kind of thing peer review should clean up. The helium level also drops by one point six millimetres during a run through evaporation, which the paper reports but does not analyse.

Host 1And access to the data?

Host 2The datasets are available on reasonable request, not publicly, and it is a preprint. So for now, outsiders cannot check the correlation analysis themselves.

What to watch

Host 1What should we watch for next?

Host 2Repeated runs across temperatures and flow settings, which the abstract says they can tune, to see whether the steady-state amplitude and the mixing pattern are robust. A theory that derives the saturation amplitude from the wave interactions, instead of fitting around it. And measurements closer to the core, where the amplification happens.

Host 1And on the astrophysics side?

Host 2Watch how theorists of ultralight boson clouds handle self-interactions, since that decides between saturation and collapse, and so what gravitational-wave searches should look for. The paper itself cites recent work finding turbulence-like behaviour in nonlinear gravity simulations, so the question of how energy cascades near black holes is live on both sides.

Sign-off

Host 1That is the deep dive: End state of the experimental black-hole bomb, by Švančara, Solidoro, Smaniotto and colleagues, led from the University of Nottingham, a preprint on the arXiv. The link is in the show notes.

Host 2Every Friday this same feed carries the roundup episode: five new papers as trailers, and you pick which become deep dives. This has been the Hastrup Review of Science.