An interactive explainer

Sound that tears water apart.

An ultrasonic bath cleans with bubbles that grow and implode tens of thousands of times a second. Follow one from the generator to the instant it collapses. Every number here is either computed live by a physics model running in your browser, or quoted from published measurements.

Unit

P 120 H

Frequencies

37 / 80 kHz

Ultrasound

330 W eff.

Tank

300×240×200

The keys on the unit work: try sweep or kHz.

It starts as electricity.

The generator drives a high-frequency signal into sandwich-type piezoelectric transducers bonded under the stainless tank. The piezo ceramic swells and shrinks with the voltage, so the tank floor becomes a loudspeaker pushing on the liquid 37,000 times a second.

Period

27.0µs

Wavelength

41.3mm

Floor motion

≈0.33µm

Above human hearing (≈20 kHz). Wavelength = sound speed ÷ frequency: 1529 m/s in water at 40 °C. Floor motion: 330 W effective spread over the 300 × 240 mm floor as a plane wave, an estimate: roughly 1/210 of a 70 µm hair. Transducer motion on screen is exaggerated.

The bath becomes a pressure map.

Sound crosses the liquid at about 1,500 m/s, bounces off the free surface and the steel walls, and overlaps with itself. The result is a standing wave: fixed zones where the pressure swings hardest, half a wavelength apart, 20.7 mm at 37 kHz.

Elma's manual notes that intensity peaks form directly above the transducers, with weaker zones towards the tank edges. Sweep keeps shifting those pressure maxima, so large parts get a more even treatment. Switch to the time-averaged view and toggle it to see the difference.

2-D finite-difference acoustic model of a slice through the P 120 H tank: 300 mm wide, 133 mm of liquid (Elma's recommended ⅔ fill), rigid steel walls, pressure-release surface, floor driven above the transducers. Sweep depth and rate are illustrative; Elma doesn't publish them.

Once a cycle, the liquid is pulled apart.

One pressure maximum, up close. The 41 mm wave dwarfs this view, so the whole liquid rises and falls in pressure together. On each low swing it can drop below zero, putting the liquid in tension. Micron-sized gas pockets are the weak points: pull hard enough and one balloons.

Elma calls these stable and transient cavitation. Threshold: Blake (1949), surfactant solution at 40 °C. Model: Gilmore + Toegel–Lohse, computed live.

Then the pressure comes back.

Radius

—µm

Wall speed

—m/s

Gas temp.

—K

Pressure inside

—bar

Water vapour

—%

Slowed

—×

Near a surface, the collapse turns into a jet.

A bubble next to a wall can't collapse evenly. The side facing open liquid caves in first and drives a jet of liquid through the bubble into the surface. High-speed photography puts these microjets at around 100 m/s and faster. A shock wave spreads from the collapse at roughly the speed of sound. That one-two, plus the steady microstreaming around oscillating bubbles, is what lifts soil out of pores, threads and blind holes.

A spherical bubble model can't resolve jets, so this sequence is an illustration built from published high-speed imaging (Lauterborn & Bolle 1975; Philipp & Lauterborn 1998). Regime boundaries are approximate.

Now multiply by the whole bath.

At any instant, bubbles are growing and collapsing throughout the high-pressure zones. Each event is tiny and over in microseconds, but there are vast numbers of them and they reach anywhere the liquid reaches: inside blind holes, capillaries and threads that a brush can't touch.

Set up the bath yourself.

…

Reference bubble: 5 µm nucleus at a pressure maximum of 1.30 bar (100 % power). Blake threshold 0.96 bar; vapour pressure 7.4 kPa at 40 °C. “Degassed” is modelled as 40 % air saturation, an assumption for illustration. Elma: 37 kHz for coarse soil, dissolving, mixing and degassing; 80 kHz for capillaries and quieter, gentler cleaning.