Ultrasonic coffee brewing cut cold brew from 24 hours to 3 minutes

Cold brew is a waiting game. Grounds sit in water for the better part of a day, and the only lever a coffee bar really has is how far ahead it remembered to start. A 2024 paper in Ultrasonics Sonochemistry describes a machine that deletes the wait. The team bolted an ultrasonic transducer to the side of an espresso machine and turned the coffee basket itself into a reactor. What normally takes 24 hours came out in under 3 minutes. The speed is not the interesting part. What the trained tasters could still tell apart is.

A 28 kHz transducer bolted to an espresso machine

The hardware is less exotic than it sounds. A Breville Dual Boiler BES920 was taken apart and fitted with a bolt clamped Langevin transducer running at 28 kHz, driving a shaped metal horn. The portafilter was cut with a window in its side wall so the horn could reach through and press against the outside of the filter basket. Sound enters the coffee through metal, not through a probe dipped into the liquid.

That detail carries more weight than it looks. Because the transducer never touches the water or the grounds, the authors note that the brewing process remains free from contamination, which is the standard objection to ultrasound anywhere near food. The full method and every measurement were published open access. The basket was an ordinary two cup Breville unit rated for 18 g of coffee.

What the sound does to a wet bed of grounds

Ultrasound at this power does not heat the coffee into submission. It makes bubbles. Acoustic cavitation grows and collapses voids in the liquid thousands of times a second, and every collapse flings liquid at whatever sits nearby. Injected through the walls at several points at once, the basket stops being a container and becomes a sonoreactor, with mixing zones spread right around it.

The second effect is quieter and probably does more of the work. Sound moving through a liquid drags the liquid along with it, a steady drift the paper calls acoustic streaming, and drift means fresh water keeps arriving at the surface of every particle. A jar in the fridge has only diffusion for that. Diffusion is why it needs a day.

The extraction yield doubled, which is not the same as improved

At 100 W, with the basket filled only a third of the way, the extraction yield climbed from 15.05% to 33.44%. Caffeine roughly doubled as well, 0.91 mg/mL to 1.84 mg/mL at a two thirds fill. Total fatty acids went from 1.16 mg/mL to 9.20 mg/mL, an eightfold jump that says something uncomfortable about how much oil a long cold soak leaves sitting in the grounds.

A number that high is a flag, not a trophy. The familiar target for drinkable coffee sits far below it, and anyone who has read what the 18 to 22 percent window means knows how pushing past the top of it tends to taste. The paper measures that figure. It never recommends it. Emptier baskets pushed it higher still, which says the limit was access to the grounds rather than anything about the beans.

Eleven tasters, eighteen attributes, thirteen matches

Physics is one thing. Eleven trained panellists were handed three cups: one sonicated for 1 minute, one for 3 minutes, and a control that had steeped at 4 °C for a full 24 hours. They scored 18 attributes each, from aroma through to aftertaste.

On 13 of those 18 the fast cups and the slow one could not be told apart. The five that separated them were aroma intensity, dark chocolate aroma, sourness, bitterness, and bitterness in the aftertaste. The 1 minute cup read as slightly under extracted and the 3 minute cup as over extracted, a shape that will be familiar to anyone who has followed what cold water changes and what it does not. The authors settle on somewhere between the two.

The one lever that also works on a jar

Most of this is shop equipment. Ultrasonic coffee brewing needs a transducer tuned to resonate through one particular basket, which is not a thing a kitchen improvises, and the paper frames it plainly as a way for cafes to stop handing over fridge space to vessels that sit overnight.

One finding travels anyway. When the team swept the brew ratio from 1 to 18, yield peaked at a ratio of 10 and then flattened out, for the sonicated and the untouched samples alike, 19.10% against 14.73%. Temperature behaved the same way. The advantage narrowed as the water warmed and disappeared at 90 °C, where sound had nothing left to offer that heat had not already delivered. At 22 °C the transducer was buying what a kettle gives away.

Worth trying on the next batch: leave the vessel loosely filled instead of packed, run it at a ratio of 10, and taste it beside the jar that has been sitting since yesterday. The grind stays put and so does the brewing time. The only thing that changes is how much room the water has to move.

Which beans are worth a three minute brew?

Speed only changes how quickly water reaches what is already in the grounds. It cannot put back what the roast left out, and a bag open since spring will simply surrender its dullness faster. That is the thinking behind the Santo Café pouch of Mexican arabica: freshly roasted, sized to be finished rather than stockpiled. Three minutes or a day, the beans decide what turns up in the glass.