# Why do we heat salt water to desalinate it rather than freeze it?

Most sea water desalination plants rely on the basic principle of heating water to evaporate it, after which the salt-free vapor is condensed. According to Wikipedia, over half the cost of desalination is made up of energy cost, since heating water is very energy intensive. As a back-of-the-envelope estimate, the energy required to bring $$1~kg$$ of room temperature ($$20$$°) to a boil under atmospheric pressure is about $$620~kcal$$. $$80~kcal$$ to heat it to $$100$$°C and another $$540~kcal$$ to bring it to a boil.

Conversely, cooling it down to $$0$$° takes $$20~kcal$$ and $$80~kcal$$ to freeze it, in total $$100~kcal$$, less than $$1/6$$ of the energy requirement of the heating method.

This kind of desalination procedure happens in nature - icebergs are largely frozen fresh water. Nevertheless, it seems this is not common in commercial desalination.

Is there a physical reason why freezing water is less suited for desalination compared to heating it?

• Icebergs are made from precipitation, coming from evaporated water. They do not freeze out of the salt water oceans. Freezing a concentrated brine will, ultimately, result in a portion of the solid with less salt concentration. You would then need to separate that piece out, and it cannot be refined further (you are at the solubility limit). Evaporation, since the salt does not go into the gas phase, is a process that can be run continually to keep making distilled water. May 16, 2019 at 13:43
• However, both methods can be used to separate alcohol... @JonCuster you should make this an answer.
– user207455
May 16, 2019 at 14:16
• The energy computation for cooling in the OP is not correct. The $100 kcal$ is the amount of heat that must be removed from the water (that number is only an approximation, by the way, since the water is presumed not to be pure). But removing heat from water requires energy, otherwise we wouldn't need electricity to run our freezers. May 16, 2019 at 14:41
• @RickGoldstein I am aware that this is energy that is leaving the system. If we used a heat pump with $\eta = 1$, we would need to expend that amount of energy. Since this is in reality more like $\eta = 0.4$, it would rather be $250~kcal$, which is still only about half of the energy for boiling the same amount of water. May 16, 2019 at 15:46

Looking at a phase diagram for water (this one By Materialscientist - Own work, CC BY-SA 3.0, Link) one sees that, starting below 23.3 wt% NaCl in water and above 0C you have a liquid solution (the oceans are around 3.5 wt% salts, mostly but not all NaCl).

So, what happens as you cool it? Well, the liquid solution cools at that NaCl content in water, until it hits the green line. Cooling further results in formation of ice containing the solubility limit of NaCl in ice. This value is pretty low, so lets assume it is, in fact, pure ice. (Note that other salts in the sea water will have different phase diagrams and solid solubilities in the ice.)

Continuing to cool the sample results in more formation of "pure" ice, leaving an increasingly saline liquid solution, requiring the sample to be cooled further. At some point one finally reaches the eutectic point at about -21.1C and 23.3 wt% NaCl. Cooling further results in formation of a ice/NaCl mixture. No further separation is possible. Now you can unload the chunk of solid, cut off the 'pure' ice, and melt that to use as water. Throw the rest away.

So, yes, you can use the freeze method (assuming that the resulting ice from real sea water meets your specs) in a batch process. Don't forget in an energy balance that you need to account for cooling of the solution to -20C or so.

Distillation can be done in a continuous flow process, which (ask a local chemical engineer) is much more desirable as a production method. Note that icebergs are the result of a just such a distillation process, using natural evaporation from oceans and other water sources to ultimately produce snow fall which then generates ice.

• Words fail me, superb... plus 1.
– user207455
May 16, 2019 at 17:24