Why does salted ice get
colder than 0°C?
Ice sits at 0°C. It shouldn't get any colder than that. Yet just sprinkling on salt can drop it to nearly −20°C — no freezer, no electricity involved. And this is exactly the same phenomenon as the salt spread on winter roads to melt ice. How can the same salt used for melting also be used for cooling?
Put ice in a cup and stick in a thermometer. After a while, the reading settles at 0°C. Even on a hot day, as long as ice remains, it stays at 0°C.
Now add a handful of salt and stir. The number on the thermometer starts sliding down. −5°C, −10°C. Done well, it can fall to nearly −20°C.
Salt isn't cold — it was sitting at room temperature in the kitchen. Nothing cold was added, yet the whole mixture got colder.
Where did the heat go? The answer: it was used up melting the ice.
Salt water doesn't freeze at 0°C. So salted ice starts melting even below 0°C. It keeps melting at temperatures where it normally wouldn't.
Turning a solid into a liquid takes a lot of heat. That heat is pulled from the surroundings. As long as melting continues, heat keeps being drawn away.
These two combine. Lowering the melting temperature keeps the ice melting, and that melting keeps pulling away heat. So the temperature keeps dropping. Let's go through it step by step.
Why does ice stop at 0°C in the first place?
Turning a solid into a liquid takes a huge amount of heat. Melting one gram of ice takes about as much heat as it takes to warm that same gram of water from 0°C to nearly 80°C.
That heat is pulled from the surroundings. That's why a drink with ice in it gets cold. Ice doesn't cool things because it's "cold" — it cools things because it's "melting."
But ice can only melt up to 0°C. Once it hits 0°C, it stops melting. If it stops melting, it stops drawing heat. So the temperature holds at 0°C. The 0°C wall is really a "can't melt any further" wall.
Salt shifts that wall
Salt water doesn't freeze at 0°C — the same reason seawater struggles to freeze even in the depths of winter. Dissolve something in water, and its freezing point drops.
When you sprinkle salt on ice, it dissolves into the thin film of water on the ice's surface, forming a strong brine. That brine can stay liquid not just at 0°C but even at −10°C.
From the ice's point of view, this means it's still "allowed" to melt. It melts at −5°C. It melts at −10°C. As long as it keeps melting, heat keeps getting drawn away.
So the temperature keeps falling. It doesn't fall forever — it stops once the brine itself finally freezes. For table salt, that limit is said to be around −21°C.
You'll often see the explanation that "salt cools things because dissolving it absorbs heat." This isn't the main reason.
The heat exchanged when table salt dissolves in water is tiny — nowhere near enough to explain a drop to −20°C. What's really drawing away the heat is the melting of the ice. Salt's role isn't to absorb heat — it's to create the conditions that let the ice keep melting.
There's proof, too. Put salt alone into water and it barely cools at all. Without ice, this phenomenon doesn't happen.
It's just telling the ice, "keep melting."
The same salt "melts" ice on the road
Here's a puzzle. Winter road de-icer is also salt. That's spread to melt ice, while this is used to cool something. It looks contradictory.
But exactly the same thing is happening. All salt does is "lower the freezing point." The difference is just what you're using it for.
At −5°C, ice normally wouldn't melt. But sprinkle on salt and it starts melting even at that temperature. The ice clears from the road surface — goal achieved.
The same thing is happening, but here the focus is on the heat being drawn away. Melting cools the surroundings, and that's what gets put to use.
In other words, "melting" and "cooling" are just two sides of the same phenomenon. Whenever ice melts, its surroundings get colder — every time, without exception. On the road too, the area right around the salt briefly cools down. It's just that nobody notices, or cares.
- It stops working if it's too cold. Table salt hits its limit around −20°C. In colder regions, calcium chloride and similar compounds, which work at lower temperatures, are used instead.
- Calcium chloride releases heat as it dissolves. Unlike table salt, this actually helps melt the ice too.
- Meltwater can refreeze. If the diluted brine refreezes in shade or overnight, it forms a clear, hard-to-see ice film. Salt having been spread doesn't guarantee safety.
- It affects metal and plants. Known effects include corrosion on the undersides of vehicles, deterioration of concrete, and damage to roadside soil and trees. Efforts continue to minimize the amount used.
Something you can check in your own kitchen
- Into a small zip-top bag, put about 100 mL of milk and 1 tbsp of sugar, press out the air, and seal it tightly
- Into a larger zip-top bag, put plenty of ice and 3–4 tbsp of salt
- Place the small bag inside the large bag, then seal the large bag too
- Wrap it in a towel (to protect your hands from frostbite) and shake continuously for 10–15 minutes
- Once the contents have set, you're done. Measure the temperature of the outer bag before and after shaking, and you'll find it has dropped well below freezing
You get ice cream without ever using a freezer. Try the same thing without salt, and the temperature only reaches 0°C, so it never sets. Trying both side by side makes the difference obvious. Don't hold the bag bare-handed for long (it can approach −20°C, risking frostbite). Don't drink the salt water — just pour it down the drain.
Summary
Salted ice gets cold not because the salt itself is cold, nor because dissolving absorbs heat. It's because salt lowers the "freezing point," keeping melting going when it would otherwise have stopped, and that ongoing melting keeps drawing heat from its surroundings.
Melting ice on the road, and cooling ice cream.
They're the same phenomenon, seen from opposite sides.
Want to go deeper? ― terms, formulas, and how this connects to the textbooksFrom middle-school science up to topics still being researched — each level is labeled
- Middle schoolCovered in middle-school science
- High schoolCovered in high-school "Basic Chemistry" / "Basic Physics"
- High school +Covered in high-school "Chemistry," or treated as advanced/sidebar material in textbooks
- UniversityNot covered in high school — a university-level specialist subject (physical chemistry)
- ResearchNot yet settled even at university level — something researchers are actively investigating
Middle schoolTerms: the vocabulary of cold
- Latent heat of fusion: the heat needed to turn a solid into a liquid. For ice, it's about 334 joules per gram — equivalent to the heat needed to warm the same amount of water by about 80°C.
- Freezing-point depression: the phenomenon where dissolving something in water lowers its freezing point. This is the real identity of the "shifted wall" in the main text.
- Freezing mixture: a combination of materials — like ice and salt — that produces low temperatures when mixed. These have been used since before freezers existed.
- Eutectic point: the limiting temperature below which salt water can no longer be cooled further. For salt water, this is said to be around −21°C, at a salt concentration of roughly 23% by mass.
- De-icing / anti-icing agents: chemicals spread on roads, including sodium chloride (table salt), calcium chloride, and magnesium chloride.
High schoolChecking it with a formula: why salt and not sugar?
Everyone knows "salt on ice," but the question of why sugar won't do rarely comes up. A calculation settles it.
ΔT ＝ K × m × i
| ΔT drop in freezing point | units: °C |
| K constant fixed by water | 1.86 (per kg of water) |
| m amount dissolved | units: mol (per kg of water) |
| i number of pieces it splits into in water | 2 for salt, 1 for sugar |
Notice that the formula counts "number of particles," not "weight." What matters isn't how many grams you added, but how many particles ended up scattered through the water. What kind of particle doesn't matter.
And that's where i matters. Salt splits into sodium and chlorine in water, so one particle becomes two. Sugar doesn't split, so it stays as one.
| Weight of 1 mol of salt | about 58.5 g |
| How many mol is 100 g? | 100 ÷ 58.5 ≒ 1.71 mol |
| Number of pieces | i ＝ 2 |
| Plug into the formula | ΔT ＝ 1.86 × 1.71 × 2 |
| Answer | ΔT ≒ drops 6.4°C |
| Weight of 1 mol of sugar | about 342 g |
| How many mol is 100 g? | 100 ÷ 342 ≒ 0.29 mol |
| Number of pieces | i ＝ 1 |
| Plug into the formula | ΔT ＝ 1.86 × 0.29 × 1 |
| Answer | ΔT ≒ only drops 0.54°C |
The difference is 6.4 ÷ 0.54 ≒ 12-fold — from the same 100 g.
Two things in the formula explain it. First, sugar molecules are heavier, so the same 100 g contains only a sixth as many particles. Second, sugar doesn't split in water, cutting the effect in half again. Multiply those together and you get roughly the 12-fold difference.
It's not that "salt is good" — it's that "something light, which splits apart in water, is good." You'd never arrive at that rephrasing without looking at the formula.
Time for some honesty. Real freezing mixtures reach around −21°C. That's nothing like the −6.4°C from step ②.
The formula isn't wrong — it's just being used outside its range. It's an approximation that holds for dilute solutions; as the salt gets more concentrated, the particles start interfering with each other and the prediction drifts off. The formula also can't explain why adding more and more salt eventually stalls out around −21°C (that's the eutectic point, covered in the next section).
| Calculated value for 100 g salt / 1 kg water | about −6.4°C |
| Measured value at the maximum amount of salt | said to be about −21°C |
Every formula has a "range where it applies." Learning to ask, when it doesn't match reality, "is the formula wrong, or am I outside its range?" is what turns calculation into a usable tool.
| Latent heat of fusion for 100 g of ice | 334 × 100 ＝ 33400 J |
| Heat to cool 100 g of water by 1°C | 4.2 × 100 ＝ 420 J |
| How many degrees' worth is that? | 33400 ÷ 420 ≒ about 80°C worth |
※ These are representative approximate values. This is where the difference shows up: the star of the cooling effect is "melting," not "salt dissolving."
High schoolHigh school +Why is it harder to freeze when something's dissolved in it?
Water freezing means water molecules lining up in an orderly crystal. If other particles are mixed in, they get in the way of the water molecules trying to line up. As a result, you have to cool it further before a crystal can form.
What's interesting is that the size of this effect depends not on "what" is dissolved but on "how many particles" are dissolved. The same number of particles of sugar or salt produces the same drop. Properties like this are called colligative properties.
This is where salt has the edge. NaCl splits into Na⁺ and Cl⁻ in water, so the same amount produces twice as many particles. Sugar doesn't split, so it stays as one. Salt is used as a de-icer partly because it's cheap, and partly because of this efficiency.
High school +Written as a formula: ΔTf = Kf × m × i. Kf is about 1.86 K·kg/mol for water, m is the molality, and i is the number of particles produced on dissociation (ideally 2 for table salt).
UniversityWhy there's a limit ― the eutectic point
Adding more and more salt doesn't keep lowering the temperature forever. As you cool salt water, ice starts separating out first, concentrating the remaining liquid. The more concentrated it gets, the further the freezing point drops — but at a certain concentration and temperature, ice crystals and salt crystals start forming at the same time. That point is the eutectic point: for salt water, around −21.1°C, at a salt mass fraction of roughly 23%. Beyond this, the whole mixture solidifies, so no further cooling effect is available.
Also, the formula above, ΔTf = Kf·m·i, is only accurate for dilute solutions. In concentrated salt water, ions interact with each other and the effective i becomes smaller than 2. A rigorous treatment requires the concept of activity, discussed using phase diagrams.
ResearchWhat's still unsettled
- The surface of ice has a thin liquid-like layer even below 0°C. This is called surface melting (or premelting), first proposed by Faraday in the 19th century. Its existence is well established, but exactly how its thickness changes with temperature, and how far you can call it "liquid," produces different results depending on the experimental method, and research is still ongoing.
- "Why is ice slippery?" has long gone unresolved. The old explanation was "pressure melts it," but calculations show the pressure from a skate blade barely lowers the melting point at all, so this explanation is now considered insufficient. Both frictional melting and the surface liquid layer mentioned above are thought to be involved, but their relative contributions shift with conditions, and no unified explanation exists yet. Despite how much sliding happens every winter, the reason for it still isn't settled. For more, see our article on why frozen roads are slippery.
- Living things use a "different trick" to prevent freezing. Antarctic fish and some insects carry antifreeze proteins. Rather than lowering the freezing point through concentration, these work through an entirely different mechanism: attaching to the surface of ice crystals and blocking their growth. The details of how they work, and how to reproduce them artificially, are still under study, with hopes for applications in food freezing and organ preservation.
- Assessing the environmental impact of de-icing salt is also ongoing. Rising soil salinity, runoff into groundwater and rivers, and effects on roadside trees have been reported in various regions. Techniques to reduce the amount used, and alternative agents, are being developed, but balancing cost and effectiveness isn't simple.
Connections to the textbooks (by level)
| Level | Subject / unit | Where in this article |
|---|---|---|
| Middle school | Science: states of matter and heat / dissolving | Why melting needs heat; why ice stops at 0°C |
| High school | Basic Chemistry: amount of substance / Basic Physics: heat quantity and specific heat | Calculating latent heat of fusion; the comparison to "about 80°C worth" |
| High school | Chemistry: properties of solutions (freezing-point depression) | Colligative properties; why an electrolyte doubles the particle count |
| High school + | Chemistry: colligative properties of dilute solutions | The formula ΔTf ＝ Kf・m・i |
| University | Physical chemistry: phase equilibrium | Eutectic point, phase diagrams, activity, where the formula stops holding |
| Research | Surface science / cryobiology (unresolved) | The liquid layer on ice surfaces, why ice is slippery, antifreeze proteins |
- Atkins, P. & de Paula, J., Physical Chemistry (freezing-point depression, colligative properties, phase diagrams and the eutectic point).
- Dash, J. G., Rempel, A. W. & Wettlaufer, J. S., The physics of premelted ice and its geophysical consequences, Reviews of Modern Physics 78, 695–741, 2006 (surface melting of ice).
- Rosenberg, R., Why is ice slippery?, Physics Today 58(12), 50–55, 2005 (explains why the pressure-melting theory is insufficient).
- Materials from Japan's Ministry of Land, Infrastructure, Transport and Tourism (国土交通省) and various road authorities on de-icer spreading and environmental impact.
- Davies, P. L., Ice-binding proteins: a remarkable diversity of structures for stopping and starting ice growth, Trends in Biochemical Sciences 39(11), 548–555, 2014 (antifreeze proteins).
※ Figures such as the latent heat of fusion and eutectic point can vary depending on conditions and measurement method. This article gives commonly cited approximate values.
※This article is a general-audience science explainer. When trying the observation for yourself, the ice-and-salt mixture drops well below freezing, so avoid touching it with bare hands for long periods. Follow guidance from your local road authority or municipality regarding the spreading of de-icing agents. The figures given are approximate values meant to aid understanding of the underlying mechanism.