A frozen-dessert science fact
Why Can Ice Cream Feel Sandy?
Not every crystal in ice cream is made of ice. In some dairy ice creams, a sugar already present in milk can form crystals of its own. When they grow large enough to notice, the sensation can be more like a fine grit than a cold crunch. Food scientists call it sandiness. There is no sand involved.
That distinction gives us a better question than “Did this get too cold?” What, exactly, is the particle you are feeling? A bit of ice, a sugar crystal, and a crumb can all interrupt a smooth bite, but they get there by different routes. The sugar story starts with something easy to overlook: even a firm scoop still has a concentrated liquid portion.
Freezing water leaves milk sugar behind
Lactose is the sugar naturally present in milk. In an ice cream mix, it can be dissolved in water, much the way a spoonful of sugar disappears into a drink. You cannot see individual grains once they are dissolved, but the sugar is still there. The University of Guelph’s guide to nonfat milk solids puts lactose in the same ingredient group as milk proteins and minerals, which helps explain why its amount is not just a matter of how much table sugar goes into a mix.
As some of the water freezes, it joins the ice crystals. Much of the dissolved material is left in the liquid between them. Imagine starting with two equal measures of water and a fixed amount of dissolved sugar. If you could move one measure of pure water out of the solution, the remaining liquid would have roughly twice as much sugar per measure of water. That is an illustration, not a recipe or a measurement of a scoop. It shows why a change can happen even when nobody adds more sugar.
In the language of the 1995 study by Yoav Livney, Daniel Donhowe and Richard Hartel, the unfrozen portion can become supersaturated with lactose. It holds more dissolved lactose than would be stable at equilibrium under those conditions. That creates an opportunity for crystals to form, but it does not tell you how quickly they will appear or how large they will get.
A crystal needs a beginning and time to grow
Scientists distinguish the start of a crystal from what happens afterward. The start is called nucleation: a small, ordered group forms, giving other molecules a place to join. Growth is the addition of more molecules to that crystal. A first crystal can be far too small to notice. Starting early and ending up large are different outcomes, so researchers can measure them separately.
There is a useful tension here. As more water turns to ice, the remaining liquid can get more concentrated. Yet as the sample gets colder and the concentrated liquid becomes more viscous, the molecules also have a harder time moving around. Picture the difference between stirring a thin drink and dragging a spoon through a thick syrup. The comparison is not a molecular model, but it gives you a feel for why having plenty of dissolved sugar nearby is not the same as getting it to a crystal surface quickly.
It is also why researchers can find solid particles with shapes worth studying. A 1930 Journal of Dairy Science report identified diamond-shaped crystals from sandy ice cream as a form of lactose containing water in its crystal structure. The shape was an observation under research conditions. You should not expect to spot tiny diamonds with the naked eye in a cone.
Colder did not always mean faster in the lab
Livney and colleagues tested one vanilla ice cream in an accelerated storage setup with controlled temperatures and temperature swings. With the temperature cycling one degree Celsius either side of the average, the first crystals appeared fastest at average temperatures between minus 10 and minus 12 degrees Celsius. Cooling further lengthened the wait again. At minus 20, the crystals grew very slowly in the experiment.
The team also found that the effect of the swings depended on which outcome they measured. At an average of minus 10 degrees Celsius, bigger swings reduced the growth rate. That is a good reminder that a general warning about fluctuating freezer temperatures does not predict every physical change in every dessert. This experiment was tracking lactose. The changing size of ice crystals is a separate question, covered in our guide to why ice cream gets icy.
Those temperatures are research conditions, not a setting to copy at home. The result is useful because the relationship was not a straight line: different conditions favored different stages, and the study distinguished when crystals first appeared from what they did next.
Sometimes the trick is to keep the crystals small
It might sound as though the only way to get a smooth texture is to prevent every crystal from forming. In a different dairy product, sweetened condensed milk, the University of Guelph describes an approach that does nearly the opposite. Its processing guide explains that small lactose crystals can be added as seeds. Many tiny starting points help keep the resulting crystals from getting so large that they make the milk feel sandy.
Think of a fixed pile of blocks. Divide it among a few builders and each can make a tall tower. Divide it among many and the towers can be smaller. Real crystals do not divide the material evenly like that, but the picture shows why encouraging more places to start can help control size. This example is about condensed milk, not a tip to stir powder into ice cream or a claim about how Sweet Escape desserts are made.
There's another lesson in T. A. Nickerson’s 1962 ice cream study. It tested explanations for why commercial samples were less often sandy and identified the gum stabilizers as the primary factor in the reduction it studied. In those tests, replacing some sucrose with corn syrup solids did not delay the start of lactose crystals or slow their growth once started. The two changes were not interchangeable. That is more useful than a general rule that swapping one sugar for another will always make ice cream smoother.
The same question comes up in dulce de leche
Frozen water is not required for this effect. In 1990, Guillermo Hough, Edgardo Martínez and Adriana Contarini compared crystal measurements with how people sensed grit in dulce de leche, a concentrated dairy sweet. In their tests, crystals below six micrometers did not register as sandy; above that size, the number of crystals mattered to detection. The team compared two kinds of evidence: what a microscope could count and what a person could feel. Those are not the same measurement. A sample can contain visible particles under magnification and still seem smooth to the panel. That figure belongs to the dulce de leche experiment. It is not a universal cutoff for ice cream or a specification for a caramel swirl.
There is a neat contrast between the two foods. In ice cream, water can leave the liquid portion by becoming ice while staying inside the scoop. In a concentrated dairy sweet, water is removed as the product cooks down. Both can leave dissolved material more crowded. What happens next depends on the food, which is why a result from a pot of dairy confection does not automatically describe the swirl in a frozen scoop.
That difference is worth keeping in mind when reading any texture study. A number tells you more when you know what food was tested and what the researchers measured. Here the cutoff described a sensation in one kind of sweet, not how much sugar it contained. A smoother bite does not mean there was less lactose in the original mixture.
What to notice in a bite
Guelph’s texture guide gives a helpful clue: an ice crystal can melt in your mouth, whereas a lactose crystal does not melt just because the ice around it does. That makes a persistent fine grit a different observation from a brief bit of cold crunch. It is a clue, not a lab test. A piece of cookie or another deliberate inclusion can also feel like a particle, and texture alone cannot tell you a food’s ingredients or handling history.
At Sweet Escape in Memphis, the Dulce de Leche entry describes caramel ice cream with a caramel swirl. That name is a fun connection to the study, not evidence of how the swirl was made, what amount of lactose it contains, or whether it forms crystals. The listing declares milk and notes equipment that also processes peanuts, tree nuts, wheat, soy and eggs.
Pick a scoop, then notice the texture
Browse the scoop catalog for descriptions and check the live menu before visiting us in Memphis. If ingredients, nutrition or allergens matter to your order, start with the item’s published label and ask the shop to verify the current container and relevant facility, shared-equipment, handling and cross-contact information.