Directional freezing, explained

Ice is a crystal lattice, and a lattice is fussy about what it will accept. As water freezes, the growing crystal takes water molecules and rejects almost everything else - dissolved oxygen and nitrogen, calcium and magnesium salts, chlorine compounds, fine particulates. Those rejected solutes do not vanish. They are pushed into the water that has not frozen yet, which becomes progressively more concentrated as the freeze proceeds.

That rejection is the whole story. Cloudy ice is not dirty ice and it is not a sign of bad water. It is ice that froze so fast, and from so many directions at once, that the rejected material had nowhere to be pushed to and got trapped as microscopic bubbles and inclusions. Those bubbles scatter light, and scattered light reads to your eye as white.

So there are two levers, and every clear-ice product on the market pulls one of them: control the direction the freeze travels, or stop air being dissolved in the water in the first place.

1. Insulate every face but one. Foam, a cooler, a thick-walled shell - anything that slows heat leaving the sides and bottom.
2. Leave the top open to the freezer air. That face loses heat fastest, so the freeze begins there.
3. Let it grow downward. A single freezing front travels from the top of the water column to the bottom.
4. Watch the impurities get pushed ahead of it. Air and minerals concentrate in the shrinking pocket of liquid below.
5. Stop before the bottom freezes, or discard the bottom. The last water to freeze is where everything you rejected ended up, so it freezes white.

That is it. The cooler-in-the-freezer trick, an insulated cube tray and a foam sphere mold are the same idea at three price points. The reason a mold has a sacrificial base, or a pour-off step, is that the method does not remove impurities - it relocates them, and something has to be thrown away.

This is also why directional freezing takes so long. You are deliberately slowing the freeze down and forcing it through one face only. A cycle of 18 to 24 hours is not a flaw in the product; it is the mechanism working. How to make clear ice walks through a single cycle step by step.

A countertop clear-ice machine does not insulate anything. It freezes water in very thin layers while the water is moving - sprayed, flowed or circulated over a chilled plate. Because each layer is thin and the water is in motion, dissolved air is carried away rather than trapped, and the sheet or cube that builds up is clear from the start.

The practical differences are worth laying out, because they decide which product is right for you rather than which is better.

Clear ice molds vs ice machines takes that table further into the ownership question. The short version: a mold is the right first purchase for almost everybody, and a machine is the right purchase for people who have already decided they want ice without planning for it.

Clarity is not purely cosmetic, though it is mostly cosmetic. Trapped air makes a cube structurally weaker and gives it slightly more effective surface area, so a cloudy cube tends to crack when you pour over it and to melt a little less predictably than a dense clear one of the same size. What actually governs dilution, though, is size and shape - a single large piece melts slower than the same mass as four small ones, clear or not.

• Clarity comes from how the water froze. Directional freezing, or a machine.
• Melt rate comes from the surface-area-to-mass ratio. One big piece beats several small ones.
• Cracking comes from internal stress, which trapped air makes worse. Tempering a cube for a minute out of the freezer helps more than clarity does.
• Taste comes from the water, not the method. Filtered water in, better ice out - and it extends the life of any machine.

If your current ice is cloudy and you want the diagnosis rather than the theory, why is my ice cloudy works through the causes in order. If you want to know which ice belongs in which drink, best ice for whiskey covers the sizing question.

1. Insulate properly, not decoratively. A thin plastic shell marketed as insulated is not the same as real foam. This is the single biggest difference between molds that work and molds that do not.
2. Do not rush the cycle. A warmer freezer setting produces clearer ice than a very cold one, because a slower freeze front rejects impurities more completely.
3. Use filtered water if you have it. It will not fix a bad mechanism, but it reduces the mineral load that has to be pushed ahead of the front.
4. Discard the sacrificial base. It is not a yield loss, it is the impurities you successfully removed. Crush it for a highball if you want it used.
5. Give the mold air. Jammed between frozen peas and a wall, a mold loses heat through its sides and the whole method breaks down.

Boiling the water twice is the advice that circulates most and helps least. It does drive off some dissolved gas, but the water reabsorbs air as it cools and the effect on a 24-hour directional freeze is marginal. Insulation and freeze rate do the work.