Sonoluminescence

How can ordinary sound in a glass of water squeeze a bubble hard enough to make it emit flashes of light?

Drive a flask of degassed water with a piezoelectric transducer at a few tens of kilohertz, seed a single small bubble, and it sits at the centre emitting a flash of light on every acoustic cycle, tens of thousands of times a second. The sound is loud but not extraordinary, and the water stays cool enough to touch.

The objection is an energy objection. Sound in water is feeble per unit volume, and light — particularly the ultraviolet-leaning continuum this bubble emits — implies matter at thousands of kelvin. Nothing in the flask is hot.

So the question is not whether sound carries enough energy, but what could take energy spread thin and slow and deliver it, intact, to a speck of gas in a few nanoseconds.

Begin with what holds the bubble still, since a wandering bubble will not do this. A standing wave has a pressure antinode at the flask's centre, and a bubble smaller than resonance feels a net force — the primary Bjerknes force — pushing it toward that antinode. The geometry traps it, which is why the demonstration uses a resonant flask.

Now follow one acoustic cycle. During the rarefaction half the pressure falls below ambient and the bubble grows — a few micrometres expanding to a few tens of micrometres is the usual order of magnitude, and I offer it as that rather than a measurement. What matters is not the size but that the surrounding water has been set moving outward, and water is heavy.

Then the pressure swings positive, the water starts inward, and inertia takes over. This step does all the work: the collapse is not the gas being squeezed by the applied acoustic pressure, which is of order one atmosphere and could never do this. It is driven by the momentum of a converging shell of liquid, accumulated over the whole expansion and delivered to a shrinking sphere — the inertial collapse Rayleigh worked out in 1917 for ship propellers, long before anyone was looking for light. The same inward-moving liquid is squeezed into an ever smaller spherical surface, so the energy is not increased but concentrated, and the concentration is geometric.

Now put a number on the consequence, because this is where the temperature stops being mysterious. Compress a gas faster than heat can leave and the adiabatic relation TV^(γ−1) = constant holds. Volume goes as the cube of radius, so T ∝ R^(−3(γ−1)). For a monatomic gas γ = 5/3, so 3(γ−1) = 2 and T ∝ R^(−2): an inverse-square law in radius. A collapse to a tenth of maximum radius therefore multiplies the temperature by a hundred, and 300 kelvin becomes about 30,000. A collapse ratio of ten is not extravagant, and the arithmetic needs no exotic physics.

At those temperatures the gas is partly ionized, and a weakly ionized plasma radiates a broad continuum, which is what the spectra show: continuous, rising into the ultraviolet, without the line structure a cool gas would give.

Here I must decline the headline numbers. Published peak temperatures and flash durations vary between research groups by more than an order of magnitude, depending on the gas, the drive amplitude, and above all on the model used to convert a measurement into a temperature. Quoting one would dress a live disagreement as a fact. Two further things are genuinely contested: whether an inward-converging shock forms inside the gas, and whether the emission is best described as thermal or as plasma bremsstrahlung. The inertial-collapse framework is not in doubt; its last stage is.

One piece remains, explaining why the effect is stable at all, since a bubble of air driven this hard should dissolve within a few cycles. The standard account — recalled as the argon rectification hypothesis, from Lohse and colleagues — is that at each collapse the nitrogen and oxygen dissociate and their reactive products dissolve into the water, while the roughly one per cent of argon in air, being inert, cannot go anywhere. Cycle after cycle the bubble purifies itself into essentially pure argon. The stability is a chemical filtration effect.

The boundary first: this collection reasons elsewhere about narrowing a distribution of atomic velocities with light, in laser cooling, and about cavitation as a nucleating disturbance in supercooling. This is neither. Laser cooling removes energy in vast numbers of tiny quantized kicks, and its point is selectivity; sonoluminescence adds no energy beyond what the sound carried and concentrates it by spherical convergence. Selection versus concentration is the fixed difference.

The bubble is also not "heated by the sound" in the sense of absorbing acoustic energy as heat. It is compressed by moving liquid, and the heating is ordinary fast-compression thermodynamics — which is why the inverse-square arithmetic works. And this is a single-bubble account: the light from a cavitating cleaning bath, many bubbles and chaotic, is messier and far less pinned down.

That gives the test. If the picture is right, emission must be locked to one phase of the acoustic cycle — the instant of minimum radius, once per cycle, with jitter far smaller than the period. That demanding prediction is what is observed. If flashes appeared at maximum bubble size, or at random phase, the inertial-collapse account would be finished.

A second test follows from the adiabatic argument: heat retained during compression is heat not conducted to the liquid, so a poorly conducting gas should shine brighter. Noble gases run from light, conductive helium to heavy, poorly conducting xenon, and the predicted brightening down the group is observed. Helium outshining xenon would refute that.

Every term the collection defines is gathered in the glossary.