Looking over our old answer (below), I realized that we said many interesting things but did not actually answer the question "why". Maybe whoever wrote the first draft didn't know the answer, and we just patched it to remove errors. Anyway, here's why.
Say you have a cup of pure water and a cup of somewhat salty water. As you lower the temperature some of the pure water starts to form ice crystals. The reason is that although the frozen water molecules, lined up into a crystal, have fewer ways to move around (lower "entropy") than the liquid molecules, they release heat when they freeze and that raises the entropy of the surroundings even more. So the net entropy goes up as the water freezes, as it always does on the way to any equilibrium state.
What about in the salty water? There's one extra term in the entropy change. The salt doesn't fit into the ice crystals. So as they form, the remaining salt is left with less room to roam around in, and thus less entropy. So you have to get the salt water even colder before you get a net entropy gain from freezing it.
It sounds like that explanation isn't special to salt- it should work for any molecules or ions dissolved in the water. And so it does.
Now for the old discussion:
If you live in a place that has lots of snow and ice in the winter, then you have probably seen the highway department spreading salt on the road to melt the ice. You may have also used salt on ice when making home-made ice cream. Salt lowers the freezing/melting point of water, so in both cases the idea is to take advantage of the lower melting point.
Ice forms when the temperature of water reaches 32 degrees Fahrenheit (0 degrees Celsius). When you add salt, that temperature drops: A 10-percent salt solution freezes at 20 °F (-6 °C), and a 20-percent solution freezes at 2 °F (-16 °C). On a roadway, this means that if you sprinkle salt on the ice, you can melt it. One way you can think about this is that the salt is so dry, so hungry for liquid water, that it actually pulls liquid water out of the ice. You can see that salt is hungrier for water than a drop of pure water would be, because salt certainly gets wet and dissolves when it's put near water. That's shows it's better at attracting liquid water than water itself is, and so it can pull water out of ice at a lower temperature than pure water can.
If you ever watch salt melting ice, you can see the dissolving process happen -- the ice immediately around the grain of salt melts, and the melting spreads out from that point. Those melted spots are actually very salty. Once the salt crystals are used up, they start to get diluted, more like ordinary liquid water, and their freezing point goes back up. If the temperature of the roadway is lower than 15 °F or so, then the salt can only melt a small amount of the ice. (If it gets below about 0 °F, it can't melt ANY of the ice.) In that case, spreading sand over the top of the ice to provide traction is a better option.
When you are making ice cream, the temperature around the ice cream mixture needs to be lower than 32 °F if you want the mixture to freeze. If you tried using pure ice around the bucket, it would warm up to 32 °F before it started to melt. It wouldn't absorb much heat in the process, so you wouldn't get ice cream. If there's a lot of salt around, the ice will melt at around 0 °F, which is cold enough. As the ice melts, it pulls a lot of heat from its surroundings, enough to cool down the ice cream. That's a key point- just warming up the ice or warming up the liquid doesn't soak up a whole lot of heat. Providing the energy to break molecules away from the ice into the liquid does soak up a lot of heat. So you want that melting process to occur cold enough to freeze the ice cream, not up at a temperature where the ice cream is still liquid.
By the way, you may wonder why you have to get the ice cream colder than 32 °F to freeze it. Just like salt water, it has molecules dissolved in the water (sugar,salts, proteins, ...) that lower its freezing point below that of pure water.
(republished on 07/25/06)
(published on 03/03/07)
(published on 11/05/07)
(published on 12/14/07)
(published on 09/08/08)