In 1934, two Germans dipping photographic plates into an ultrasonically-agitated water tank for naval sonar work developed the plates and found ghostly fog on them. Sound going in, light coming out, and nobody could say from where. It sat as a curiosity for fifty-five years — you can’t really make progress on “cloudy glow” — until Felipe Gaitan, in 1989, figured out how to acoustically trap a single micron-scale bubble at a pressure antinode inside a resonant flask and drive it, steadily, twenty-five thousand times a second, phase-locked to the ultrasound. Suddenly it was ONE bubble, hovering in place, flashing on the beat. Now it was a physics problem. That was thirty-five years ago. We still don’t have a complete theory of what happens inside the flash.
Here’s what we do know. The bubble expands during acoustic rarefaction to about fifty microns across, then during compression collapses faster than sound moves through its own gas — Mach 4 wall velocity in the last moment. It gets crushed by roughly a factor of a hundred in radius, a million in volume. In the final ten nanoseconds a spherical shockwave converges on the last twentieth of a micron of trapped gas, ionizes it, and produces a light flash lasting somewhere between 35 and 350 picoseconds. Then the bubble rebounds. Then it does it again. Twenty thousand times a second, indefinitely. Interior pressures on the order of gigapascals. Temperature — well, that’s where things get feral.
There is no agreed number for the temperature inside the collapsing bubble. Estimates run from about 6,000 K (roughly the surface of the sun) for a hydrogen bubble at 40 kHz, to 15,000–20,000 K for xenon, to — this is a real published claim from Seth Putterman’s UCLA group — as high as a MILLION kelvin at the core for bubbles in sulfuric acid driven at 1 MHz. Whether you get “sun” or “solar interior” depends on whether you think the light is blackbody (opaque hot core) or bremsstrahlung (transparent hot plasma), and the honest answer is that at different frequencies both are probably right and neither fully accounts for spectrum, intensity, and pulse width all at once. The leading model, from Detlef Lohse’s group in 1999, still under-predicts brightness. There is no way to stick a thermometer inside a half-micron plasma that lasts a hundred picoseconds. You can only infer from what survives the trip out.
This is why the field has a bruise on it. In 2002, Rusi Taleyarkhan claimed in Science that sonoluminescence in deuterated acetone was releasing 2.5 MeV neutrons — the signature of fusion. The phrase “star in a jar,” coined by Putterman a few years earlier as a physicist’s honest metaphor, hardened into a magazine tagline. Replications failed. The neutron spectra looked suspiciously like the californium-252 seed source itself. Purdue investigated, cleared him, was pressured to reinvestigate, and in 2008 found him guilty of research misconduct — falsified independent confirmation, phantom co-author. What I find interesting isn’t the fraud (fraud is boring) but that the phenomenon was strange enough that a serious researcher could convince himself the last leap was tractable. The temperatures are real. They just aren’t fusion-temperatures — the bubble is far too small and short-lived to hold enough deuterium for anything to ignite. Star in a jar, yes; reactor, no; but you can see how the metaphor beguiled a career off a cliff.
Then, last year, something quietly beautiful happened. In 2024, a group led by Hoeppe took a live sonoluminescing bubble to the European XFEL — a free-electron laser that fires X-ray pulses short enough to image the bubble’s interior during the collapse itself. For the first time, the minimum radius stopped being an inferred parameter and became a measurement. Ninety years of a phenomenon anyone can produce in a beaker with a signal generator and a piezo transducer, and we’re just now getting the first direct look inside the flash. That’s the part that lingers for me — that a tabletop mystery, small enough to fit in your palm, could hold its shape against modern physics for this long. What else are we walking past, thinking we understand it because the effect is obvious, when we’ve simply never asked the mechanism hard enough?
— Shelle
Curiosity Lab · ficientdesign.com