Supercooled Water in the Freezer
Most recent answer: 10/22/2007
- Greg (age 26)
In the absence of impurities in the water and imperfections in the bottle, the water can get "stuck" in its liquid state as it cools off, even below its freezing point. We say this supercooled state is "metastable." The water will stay liquid until something comes along to nucleate crystal growth. A speck of dust, or a flake of frost from the screw-cap falling into the bottle are enough to get the freezing going, and the crystals will build on each other and spread through the water in the bottle.
Water releases 80 calories per gram when turning from a liquid to a solid. We suspect your freezer is only a few degrees Celsius below zero(perhaps ten or fifteen?), and the specific heat of water is one calorie per degree per gram. This means that your water, as it freezes, warms up the rest of the water until the process stops at 0 degrees Celsius, freezing perhaps ten or twenty percent of the water. This ice may be distributed throughout the bottle, though, as the crystallization process happens very quickly and heat flows slowly.
We suspect you have slush in your bottle rather than hard ice when this is done. You can compare with another bottle which froze hard in your freezer overnight how hard it is to squeeze the bottle and how long it takes to melt. The ice will also take up more room than the water it used to be, and some water may spill out the top.
There can also be some small effects of pressure and of dissolved gases on the freezing temperature. Is your water under pressure?
Tom and Mike
(published on 10/22/2007)
Follow-Up #1: Supercooled water in the freezer
- Jordan Canevari 11
(published on 10/22/2007)
Follow-Up #2: supercooled water
- Greg (age 24)
(published on 07/02/2008)
Follow-Up #3: supercooling observed
- Eric Ford (age 37)
Mesa, Arizona, USA
(published on 08/15/2008)
Follow-Up #4: pressure and supercooling
- garrett (age 26)
(published on 10/06/2008)
Follow-Up #5: bubbles in supercooled water
- David (age 29)
Laguna Niguel, CA
It's just possible that the bubbles themselves started nucleating some freezing, so perhaps they were dragging little ice crystals along, but that's just a wild guess.
Not only was that guess wild, it was kind of stupid. See the "viscosity" follow-up below, for a much more plausible answer from one of our readers.
(published on 10/14/2008)
Follow-Up #6: desalinated water freezing
- Evan (age 34)
Thanks for the interesting info.
(published on 08/23/2009)
Follow-Up #7: Iraqi supercooled water
- Matthew O'Farrell (age 26)
El Paso. TX
(published on 11/03/2009)
Follow-Up #8: supercooled soda
- Jason (age 23)
(published on 11/03/2009)
Follow-Up #9: more data
- Craig (age 60)
Greenville, NC, USA
(published on 11/03/2009)
Follow-Up #10: more ice data
- Dan (age 24)
Rockford, IL, USA
On the theory issues: The water molecules are moving around very quickly on their own even before you pick up the bottle, thanks to thermal energy. However, they do so without any collective structure large enough to start the freezing. When you pick up the bottle, some larger scale disturbance (e.g. a small bubble) is introduced, and that can be enough to nucleate the ice.
(published on 11/11/2009)
Follow-Up #11: gas in soda
- Dave (age 19)
(published on 11/19/2009)
Follow-Up #12: Navy weighs in
- Chris (Navy) (age 26)
(published on 10/29/2010)
Follow-Up #13: more supercooling data
- Adam Houston (age 39)
(published on 11/09/2010)
Follow-Up #14: supercooled Aquafina water
- Cody S (age 26)
We're very lucky here in Urbana. The tap water is delicious, clean, costs about $0.005/gallon, and doesn't fill up the landfills with plastic.
(published on 03/10/2011)
Follow-Up #15: supercooling under pressure
- James (age 22)
I'd be surprised if you found a sharp pressure threshold for super cooling. Usually even tap water will supercool a bit under atmospheric pressure. If you do find that the amount of supercooling you can get does depend a lot on pressure, here's a possible reason.
Maybe the main nuclei are little bubbles. If you pressurize the water, you can shrink or eliminate them. Then you'll just be left with dust, etc. for nuclei. So you might find that the amount of supercooling you can get drops for the first bit of pressurization, then levels off for further pressurization. The reason that atmospheric pressure would be special is that the air dissolved in the water has had a chance to reach equilibrium with the atmosphere, so increasing pressure leads little bubbles to dissolve.
Since you find that taking the pressure off usually leads to freezing, I guess the bubbles have shrunk under pressure but not completely gone away. Maybe the one time that releasing the pressure didn't cause freezing, you'd actually managed to fully remove the bubbles.
Slightly depressurizing the water for a while, then letting it come back to atmospheric pressure without shaking it should remove bubbles. That too should increase supercooling.
You can see I'm doing a lot of guessing, but these are all testable ideas. One tool that might help would be a laser pointer. In a dark room, you can see how much the water scatters light. That scattering mostly comes from bubbles, dust, etc.
(published on 05/07/2011)
Follow-Up #16: supercooled slush and latent heat
- Becky (age 20)
San Antonio, TX
Great observation and great question.
When the freezing starts, the water molecules joining the ice lose energy. That's the whole reason they join the ice, the bonds between them in the ice are lower energy than the looser bonds in the liquid. That lost intermolecular bond energy (called latent heat) has to go somewhere. It heads straight into the vibrations of the nearby molecules, heating them up. That keeps happening until the water is back up to 0°C, at which point the freezing stops.
We've discussed this a little on some previous answers. The latent heat is big enough that for practical supercooling it will stop the freezing well before 100% of the water has frozen. That's why you end up with a slush.
(For purists, yes we should be talking about enthalpy rather than energy, but there's not much difference here.)
(published on 05/10/2011)
Follow-Up #17: supercool slurpies
- Angel (age 32)
You don't need pressure.
As for whether it can happen with any liquid, the answer is basically yes. However, the ability to supercool is bound to vary among different liquids.
I think the liquids you're interested in here are almost all solutions of things in water. The big factor that varies among them is the presence of little bits of dust etc. that serve as freezing nuclei and prevent things from supercooling very much. It sounds like most soft drinks are filtered in a way that gets rid of the freezing nuclei. The sugar, artificial sweeteners, etc. are present as individual small molecules. It doesn't sound like they play all that much of a role.
(published on 07/06/2011)
Follow-Up #18: thumping supercooled water
(published on 07/12/2011)
Follow-Up #19: Supercooled water project
- Margareth (age 46)
Gainesville, FL, United States of America
If you could get your hands on some little proteins from Antarctic fish, you could actually suppress freezing- these fish use the proteins to live in a supercooled state. They may be available since some ice-cream makers use them to get a smooth texture.
(published on 09/12/2011)
Follow-Up #20: pesticide impurities in slushy ice-water?
- Stuart Long (age 50)
Raleigh, North Carolina, USA
Let's say that, doing that, you're still having trouble getting certain water to freeze solid in a certain time. That may be because:
1. There are a lot of solutes in the water
2. The freezer isn't very cold.
The important solutes under normal conditions are standard mineral salts. One thing that almost certainly is not involved is pesticides. The reason is that even highly dangerous levels of pesticides involve rather low molecular concentrations. You'll never notice their effects on the water freezing, because those effects are negligible compared to the effects of small variations in salt levels.
It's possible that in some of the countries you visit the freezers aren't as cold as they should be.
(published on 11/08/2011)
Follow-Up #21: possible supercooling?
- Eric (age 29)
Denton, TX , USA
Some of the items you mentioned (e.g. applesauce) should be hard to supercool much, because they contain particles that would nucleate the freezing. When you say that they turn to slush, I'd guess you must mean that they're already slush when you remove them from the fridge. The reason they form slush rather than solid ice would be a little different than the reason that supercooled liquids turn to slush as they're removed from the fridge.
The supercooled liquid releases heat as it starts to freeze. That heat keeps it from staying cold enough to freeze completely. What's probably going on with your liquids is that, as they partially freeze, the sugar concentration in the remaining liquid goes up. As a result the freezing point of the remaining liquid goes down, below even the temperature that your malfunctioning fridge is reaching.
(published on 11/09/2011)
Follow-Up #22: more supercooling data
Thanks for your data and for your kind remarks.
(published on 12/28/2011)
Follow-Up #23: our staff
- David Harper (age 53)
(published on 03/16/2012)
Follow-Up #24: supercooled water cooler
- Kelly (age 49)
Phoenix, AZ US
(published on 03/31/2012)
Follow-Up #25: supercooled margaritas
- Pat (age 58)
San Jose, CA USA
When you say "the mug froze" I'm guessing you mean that there was ice on its outside.
(published on 06/03/2012)
Follow-Up #26: nucleating supercooled water
- Dana (age 58)
Aurora, Indiana, USA
(published on 06/11/2012)
Follow-Up #27: surprise supercooling
- ingrid g. (age 50)
chicago, il usa
(published on 07/01/2012)
Follow-Up #28: Thank you for the thanks
- draya (age 52)
We appreciate it when we get a note like yours. It makes our efforts worthwhile.
LeeH and MikeW
(published on 08/09/2012)
Follow-Up #29: slush from supercooled water
- Martheen (age 25)
It's fun that we can use theoretical arguments to confidently conclude things like that the product must be slush, not solid ice, and then see the result in a simple home experiment.
(published on 11/07/2012)
Follow-Up #30: supercooled water viscosity
- Rodney F. (age 66)
Sacramento, CA, USA
On the effect of pressure in melting ice, I've heard that the old story we always told about the pressure from the skates melting a layer doesn't hold up. The effect just isn't big enough.
On the water-alcohol solution, our point concerned the thermodynamics of freezing, which is simpler than the viscosity effects you discuss. For such solutions the melting point always turns into a melting range. Rather than having a single freezing point below which ice is the stable phase, as the temperature is lowered the stable state has an increasing fraction of ice.
(published on 12/14/2012)
Follow-Up #31: freezing supercooled foam!
- Adelina P (age 26)
SAlina, UT, USA
Now to try to understand it, here's some thoughts.
The reason that ordinary supercooled water only forms a slush is that the latent heat it releases warms it up as it starts to freeze. If it starts at -20°C, about as cold as you're likely to be able to supercool it, the latent heat will warm it to 0°C when about 1/4 is frozen. So in order to get all the bubbles to freeze, they must have someplace else to dump that latent heat energy. I can think of two effects that would help.
1. There's a lot of cold gas nearby. The liquid in the bubble is very thin- maybe only about 10-4 cm. If the bubbles are around 10-1 cm in size, there's roughly 1000 times the volume of gas as of liquid. (This estimate could easily be off by a factor of 10.) That means that the net heat capacity (heat energy absorbed per increase in temperature) of the gas in those bubbles can be nearly as big as that of the liquid. That still doesn't sound like quite enough to soak up enough energy to let the whole bubble freeze.
2.When you open the bottle, the bubbles expand, doing work. This is described in standard thermodynamics books as the cooling on adiabatic expansion. You could probably get gas cooling of around -60° C in this process, if I calculate right. That would let the gas soak up some more heat before reaching 0°C.
Combining those two effects would be just about enough to explain what you saw. Maybe other effects are involved as well. Perhaps, for example, only some of the fizz freezes, the rest falling back into the liquid. When you look, you see only frozen fizz. That's called "post-selection", an effect well known to distort many statistical studies.
p.s. Can you try repeating this? Was that soda with sugar? Anything else we should know before trying it?
(published on 12/26/2012)
Follow-Up #32: supercool experiments
- William (age 14)
The container probably is more important. We've heard from more people with highly supercooled water in plastic bottles. Maybe that means that glass bottles are more likely to trigger ice formation, especially on those scratches that you mention, or maybe it just means that plastic bottles are more common. The most important factor that I know about is to not have little bits of dust in the water. It's not necessary that the water be distilled, but it does help a lot if it's well-filtered. By that I mean with a particle filter, not a chemical filter. The water purified by reverse-osmosis has accidentally had particles removed as part of the chemical purification process.
(published on 01/03/2013)
Follow-Up #33: shaking triggers ice formation
- apa (age 19)
(published on 01/20/2013)
Follow-Up #34: supercooled water and slush?
- Ryan (age 34)
We're a bit puzzled.
(published on 01/24/2013)
Follow-Up #35: solid ice from supercooled water?
- George (age 56)
(published on 04/08/2013)
Follow-Up #36: frozen supercooled water
- George (age 56)
(published on 04/09/2013)
Follow-Up #37: supercooling in Estes Park
- Candice (age 42)
Estes Park, CO
So for our other readers, two parts of this are easy to understand. Initially, the water a bit below 32°F is supercooled, meaning it doesn't freeze until something disrupts a little region to get the ice formation started. Finally, something does get the ice formation started ("nucleated") and the latent heat released by the freezing raises the temperature to almost 32°F.
The mystery is the step in between. Why didn't opening the bottle and taking a sip disrupt some little region enough to nucleate the freezing? I'll just make a guess. 25° F isn't way below 32°F, so just moving the bottle around may not be enough to nucleate freezing. When your son took a drink, the part near his mouth was warmed and picked up some salt, both of which would tend to keep it from freezing. Then when the bottle went in the snowbank, some little bits of something that got in could settle down into the cold part and nucleate the freezing.
(published on 04/23/2013)
Follow-Up #38: experience with supercooled water
- Mike C (age 36)
(published on 04/28/2013)
Follow-Up #39: supercooled beer
- maxime (age 19)
1. If you read the earlier parts of this thread, you can see that because of the latent heat, it's basically impossible to supercool water enough for it to freeze solid. It should turn to partly-frozen slush.
2. That limitation is even stronger for beer, because as part of it freezes the alcohol concentrates in the remainder, keeping it from freezing.
So I bet you got partly frozen beer.
(published on 04/29/2013)
Follow-Up #40: ice water turning to slush
- Jason (age 41)
(published on 05/12/2013)
Follow-Up #41: popsicle making
- Johnah Crain (age 22)
I'm not sure what the procedure is. Adding any solute (e.g. sugar) to water will lower its true freezing point below 0°C. I think you're talking about trying to actually supercool it, to keep it from forming big ice crystals even when it's cold enough to start freezing despite the sugar. One possible way to promote supercooling is by adding some of the special antifreeze that antarctic fish have in their blood. It's a sort of very small protein. () The concentration is too low to lower the true freezing point much but it binds to ice crystals as they just start to form, greatly slowing the formation of big crystals. That's how the fish can make it through the winter below the true freezing point of their blood. Some ice creams and popsicles are made using these materials, since they help keep big crystals from forming.
(published on 07/24/2013)
Follow-Up #42: supercooled water science projects
- Rob Young (age 45)
This won't be the easiest sort of science fair project, because it takes some care to make the supercooled water reproducibly. Still, there are several interesting questions that could be explored.
One, mentioned above, would to see what sorts of things are best at triggering the freezing. How cold does the water have to be for say a little piece of some rock to trigger the freezing? How cold for some reproducible little tap on the water to do it? and so forth. You could even try something where the size of the disturbance could be varied in a controllable way, maybe via little taps from different size pendulums. How does the maximum temperature at which the tap works vary with the size of the tap?
It's probably beyond the scope of what your daughter could do for the science fair, but pursuing this type of experiment far enough could lead to something systematic on the whole topic of how nucleation of the ordered phase occurs.
Which reminds me that I've got to quit answering these and get back to helping with a research paper on that very topic, although in a different system.
(published on 09/29/2013)
Follow-Up #43: supercooled spring water?
- lily (age 19)
Any water can supercool a little bit, but your guess that the unfiltered spring water may not supercool much sounds reasonable to me. Still, the label might not be right or that spring may naturally not have many good nucleation particles in it, so the only real way to tell is by experiment.
(published on 10/18/2013)
Follow-Up #44: freezing supercooled saltwater
- Rob Y (age 45)
My first reaction is that it's hard to get highly reproducible results with supercooling, as we said in #42.
For background, let's try to calculate about how much that salt should lower the freezing point. 1/2 tsp is ~2.8 gm. The weight of a mole of NaCl is ~58 gm. So that batch had ~0.05 moles NaCl. 12 oz is ~ 1/3 liter. So you had up to ~0.15 M salt solution. That would give a freezing point depression of only around 0.5°C, not enough to make a huge difference in the net amount frozen, assuming that you've gotten the water and the sieve much more than 0.5°C below 0°C.
Yet this salt seems to have made a big difference in the freezing pattern, although maybe not in the total amount frozen. The different pattern will leave different amounts of liquid stuck to the surfaces, messing up the weight comparison, as you noticed. That small amount of salt doesn't change the viscosity of water much at all, so I think it's the different crystal pattern that accounts for the different amount of water stuck.
Why does the salt change the pattern of crystal growth so much? Here's a guess. As crystals start to form, they exclude the salt. That makes a layer of extra-salty water around them, suppressing further crystal growth. (The released latent heat of crystallization has a similar effect, but it diffuses away faster.) That promotes the formation of many little crystals rather than fewer bigger ones. That's how you get that slush.
(published on 11/30/2013)
Follow-Up #45: why does supercooling work?
- Sandra (age 33)
Sacramento, CA, USA
I've moved your question to a thread on supercooling. You give us an excuse to explain the effect at a little more depth. The question is why would water stay a liquid even though it's cold enough so that the solid ice would be more stable than the liquid.
Let's start with the basics: why are the liquid and solid separate phases to begin with? Why don't they mush together on the scale of molecules?
The stable form of things is the one that minimizes the free energy (U-TS), a balance between lowering the energy (U) and raising the number of different possible micro-arrangements, measured by the entropy (S). At low temperature (T), lowering U is more important. So when water is cold, the molecules line up into the low-U crystal form, even though that lowers their S. At high T, S is more important, so they break loose and wander among more different arrangements.
Imagine there is a little region with an ice crystal contacting a little region of liquid, near the freezing temperature. The molecules on the surface of the ice have lost entropy by lining up, but not lost much energy because on one side the liquid molecules aren't lined up with them. So molecules at that surface have higher free energy than ones in either the liquid or the solid. The molecules arrange to minimize the liquid/solid surface, separating the two phases.
What happens if you start with pure liquid and cool it enough so the solid is more stable? Let's say a few molecules by accident happen to arrange in an ice-like pattern. Mostly, they're still at the ice surface. They have higher free energy than the liquid, not lower. So usually they will just roll back down in free energy to the liquid state.
This sort of pure accidental arrangement takes a very long time to get an ice crystal started that's big enough to keep growing, unless the liquid is way below the freezing point. That's why you can supercool water until some special nucleation gets the freezing going.
(published on 01/01/2014)
Follow-Up #46: supercooled wine
- Pato (age 61)
Darwin, NT, Auustralia
For starters, the freezing temperature of the wine is around -6°C thanks to the alcohol and other solutes.
See . So -20°C is not as much supercooled for wine as it would be for water. Still 14° or so supercooling seem large, especially for something with some sediment in it. The solute molecules themselves would not serve as nucleation centers. Other readers have written in of problems with wine freezing, so perhaps you have an unusual wine, maybe filtered. (See .)
So I'm not sure how weird your result is.
As for the heat of fusion, it's about the same for supercooled liquids as for ones right at the freezing point. It shrinks below the true freezing point by an amount proportional to the difference between the heat capacities of the solid and liquid. That's not typically a big deal for ordinary supercooling. The latent heat of the alcohol solution, however, is a bit different from that for pure water. (see ) I doubt that difference matters much for what you're seeing.
posted without vetting until Lee returns
(published on 01/18/2015)
Follow-Up #47: experiments on supercooling
- Finn McGinnis (age 13)
One direct possibility is to buy some particle filters and use these to remove particles from the water. Then you can reuse the same filtered water many times. For example, 0.22 micron cellulose acetate filters should remove most dust. I think you can buy these in small quantities and use them with a standard syringe. It would help to experiment with different types of bottles. Perhaps standard polyethylene would be good. It's a good idea to avoid glas because it can break as the water freezes.
You could also buy a few brands of bottled water and see which supercool best. Some already seem to be filtered and in suitable bottles, because our readers report accidental supercooling.
(published on 12/11/2016)
Follow-Up #48: minerals and supercooling
- Bethenia Dixon (age 33)
Newport, NC, USA
The minerals will slightly lower the freezing point but have very little effect on supercooling. The reverse osmosis process has the incidental effect of filtering out not just ions and molecules but also lager particles. That's what allows the supercooling to work so well, since it's those larger particles that trigger freezing.
(published on 02/05/2017)
Follow-Up #49: acetate hot ice
- Laura Griffeth (age 38)
Rigby, ID, USA
I can't think of any ordinary liquids that can't be supercooled at least a little bit. The liquid-solid transition in 3-D is first-order, which means that even looking at a small region the structure of the liquid and solid look different. That makes it hard for the liquid to find its way over to the different solid state, since in between it has to go through some unlikely state that's not either one. (I guess you could say the liquid 3He can't be supercooled, however. That's because at atmospheric pressure the lowest-temperature phase is actually liquid, not solid, due to some quantum spin effects.)
From what I've just read about "hot ice" on Wikipedia, it sounds like the hot ice heaters use supersaturated solutions. The physics behind supersaturation and supercooling are quite similar, however. Getting that first little acetate crystal started is hard, just like getting that first little ice crystal started is hard. What's different is that when the phase transition is done, water forms a crystal. The supersaturated solution forms crystals together with leftover liquid.
I think there's a general correlation between the temperature of a phase transition and the latent heat. High temperature transitions tend to involve larger latent heat. It's not a rigid relation by any means, however. In general, phase transitions in any temperature range don't even have to be first-order (although liquid-solid is), which means that they don't even have to have any latent heat.
(published on 02/23/2017)
Follow-Up #50: water didn't freeze
- John NWOKO (age 65y)
The most likely reason is that each of the other bottles had a fleck of something in it that triggered the freezing. The last bottle maybe didn't have that, so it stayed supercooled longer.
(published on 08/03/2018)