A month or so ago, my youngest was making italian ice by freezing juice in stages, and pulling it from the freezer frequently to laboriously scrape it. That got me thinking…

I wondered what would happen if we used an ultrasonic transducer on it while it was freezing. Would micro-cavitation limit ice crystal growth and produce a smooth texture without stirring? Someone has already tried this, right?

Yes, it turns out. It’s called “sonicating” and it is commonly used in commercial ice cream production for this purpose, where of course the equipment costs thousands of dollars. (It is also used for homogenization and even pasteurization.) But I didn’t find any information on doing this at home, and none of the home ice cream makers I saw advertised included sonicators, so we started to experiment.

I opened up our ultrasonic cleaner and probed the connection between the control/display unit and the driver board. There were six wires, which were clearly +5V, ground, sonicator enable (+5V on), heater enable (+5V on), and two thermistor connections. I made an extra external connection to the header so I could control it remotely.¹ (Do note that enabling the ultrasonic transducer without a load will burn out the driver circuit very quickly; I actually did this by first making the external connections, then testing with water in the cleaner.)

My youngest has an arduino uno that he got as a birthday present from a friend a few years ago, and since it’s a 5V part, it seemed like the obvious way to start controlling this. He wrote a simple program that turns on the sonicator for 5 seconds, then turns it off for a programmable delay that he can set between 5 seconds and 60 seconds using a potentiometer. Rather than try to calibrate the thermistor, we used a DS18B20 in a waterproof package to keep track of temperature, and a buzzer to alert when the temperature gets too warm (so far, from testing, -7⁰C seems like a good upper limit). We used an LCD display to show the current off time and temperature.

We put our mix into a ziploc bag, and put it in the ultrasonic cleaner tank (not using the basket, which would attenuate the sound energy) with salt, ice, and some water (up to the “min” level on the tank), with the DS18B20 sensor in the water. We couldn’t use just salt and ice like you would in a paddle ice cream maker, because it takes liquid water to transport the ultrasonic waves from the transducer to the mix in the bag. We put the cover on the cleaner, set a few ice packs on top of the cover, and then wrapped blankets around it. The blankets both kept heat out and attenuated the awful screaming noise (I and my children hear unusually high frequencies). We used ice from our chest freezer, which is colder than our kitchen freezer. We were able to keep the mix around -10⁰C.

In our 2.5L cleaner, a quart of mix was clearly the max that we could pack enough ice around, and it took at least one refresh of the ice and salt to make ice cream. We had to remove water when we added ice so that it didn’t overflow.

Our first try, with just plain orange juice, failed because we didn’t yet have the temperature sensor and warning buzzer, nor the blankets, and after it was almost frozen, I let it get too warm, and it all melted—snatching defeat from the jaws of victory. Our second try, with a normal ice cream mix, made acceptable ice cream.

We set a 60-second delay between 5-second bursts during freezing, and we had noticeable thin flakes of ice. We’ll try again with a 20-second delay and see whether it changes the texture.

This is a tasty way to experiment!


For our next try, instead of putting a quart of mix into a single quart ziploc, we made a quart of mix into 7 different flavors, each in a separate ziploc sandwich bag. We then packed all the bags into the ultrasonic cleaner with salted ice surrounding them. It took about three hours in the ultrasonic bath, with a cycle of 5 seconds on, 20 seconds off. About once each hour to hour and a half, we had to remove water and add ice and salt after the high temperature alarm sounded; the first change was sooner because the mix started out at room temperature. We used ice that we kept extra cold in the chest freezer.

The mixes with alcohol-extracted flavorings (vanilla and cinnamon) took longer to freeze. When the packs were firm, we moved them to the chest freezer for the static freeze part of the cycle.

The texture varied by flavoring. When ice crystals were present, they tended to be flat and soft even after the full static freeze, unlike the hard crunchy crystals in ice cream that has thawed and refrozen.

Of course, there was no “overrun” of entrained air.

We really like this approach, but all the ice and salt is inconvenient. Now I’m wondering if any of the compressor-based home ice cream makers with dashers could be modifed to have ultrasonic transducers?

¹Yes, I expect my warranty is void…