Winter Rainbows And Supercooled Water

Steve Estes, a coworker in the seismology lab, walked in after lunch one day last week and announced that he'd just seen a rainbow over Fairbanks. Anyone who has spent a winter here knows that rainbows just don't happen in the Alaskan Interior during the middle of December. Even though it was an exceptionally warm day for the time of year, temperatures were still below the freezing point, which should have made the combination of sunlight and water droplets that creates a rainbow impossible--or so it would seem.

But is it? Craig Bohren points out in a recent article in Weatherwise that it is a widespread misconception that water necessarily freezes at 0 degrees C (32 degrees F). We have all heard this so many times that we have come to believe it. Yet it is more correct to say that 0 degrees C is the melting point of ice, rather than to say it is the freezing point of water.

If you were to put an ice cube into a hot frying pan, the molecular arrangement of the water molecules would change rapidly from a regular, ordered crystalline array to the more disorderly liquid phase and then into freeform steam. Things don't happen quite so abruptly going the other way. In pure water, sometimes a nudge is needed to get the molecules to make the transition from a disordered to an orderly phase.

In a gas, such as steam, the molecules are in random, incessant motion more or less independent of each other. When the steam condenses into water, the molecules spend part of their time as members of clusters which are continually forming and breaking up. To make the final transition from a liquid to the lattice structure of ice, however, some kind of foreign particle--an ice nucleus--is needed to initiate freezing. Until this nucleus forms, liquid water can exist far below the freezing point. Such water is called supercooled.

In fact, as Bohren points out, liquid water can exist at temperatures as low as -40 degrees C (or F), at which point the molecules become so sluggish that there is a high probability that enough of them will get together for a sufficiently long time to form a nucleus even in the total absence of foreign matter. Lacking any outside influence, pure water can nucleate itself even at the relatively high temperatures of a refrigerator freezing compartment. But, says Bohren, the chances of this happening are about the same as for all the air molecules in a room suddenly rushing into one corner.

If, as all this makes it sound, ice really doesn't "want" to form, why is there so much of it lying around? The answer is that even supposedly "pure" water is almost always teeming with all kinds of microscopic rubbish which can serve as ice nuclei. This is why we have come to always expect water to freeze at 0 degrees C.

Not quite willing to take all this on faith, I tried freezing some distilled water droplets in our freezing compartment at home. I sprinkled them onto waxed paper so that they didn't spread out, and tried to time how long it took for them to freeze. One of the first things I found out was that opening and closing the freezer door made for less than a controlled laboratory experiment. I also learned (as I should have remembered from high school chemistry) that shaking the paper to see if a droplet was frozen only made it freeze faster.

However, I eventually found several small droplets of distilled water that remained unfrozen for a half-hour or more. That contrasted markedly with the "control" samples of tap water which seemed to freeze almost immediately. Satisfied by experimental evidence, I will now believe Steve when he tells his story of the winter rainbow over Fairbanks.