Mpemba effect: Why hot water can freeze faster than cold

Mar 26, 2010 by Lin Edwards report
Mpemba Effect. Image: James Brownridge

(PhysOrg.com) -- Scientists have known for generations that hot water can sometimes freeze faster than cold, an effect known as the Mpemba effect, but until now have not understood why. Several theories have been proposed, but one scientist believes he has the answer.

Theories for the Mpemba effect have included:

  • faster evaporation of hot , which reduces the volume left to freeze

  • formation of a frost layer on cold water, insulating it

  • different concentrations of solutes such as , which is driven off when the water is heated
The problem is that the effect does not always appear, and cold water often freezes faster than hot water.

Radiation safety officer with the State University of New York, James Brownridge, has been studying the effect in his spare time for the last decade, carrying out hundreds of experiments, and now says he has evidence that supercooling is involved. Brownridge said he found water usually supercools at 0°C and only begins freezing below this temperature. The freezing point is governed by in the water that seed ice crystal formation. Impurities such as dust, , and dissolved salts all have a characteristic nucleation temperature, and when several are present the freezing point is determined by the one with the highest temperature.

In his experiments, Brownridge took two water samples at the same temperature and placed them in a freezer. He found that one would usually freeze before the other, presumably because of a slightly different mix of impurities. He then removed the samples from the freezer, warmed one to and the other to 80°C and then froze them again. The results were that if the difference in freezing point was at least 5°C, the one with the highest freezing point always froze before the other if it was heated to 80°C and then re-frozen.

Brownridge said the hot water cools faster because of the bigger difference in temperature between the water and the freezer, and this helps it reach its freezing point before the cold water reaches its natural freezing point, which is at least 5°C lower. He also said all the conditions must be controlled, such as the location of the samples in the freezer, and the type of container, which he said other researchers had not done.

The effect now known as the Mpemba effect was first noted in the 4th century BC by Aristotle, and many scientists have noted the same phenomenon in the centuries since Aristotle’s time. It was dubbed the Mpemba effect in the 1960s when schoolboy Erasto Mpemba from Tanzania claimed in his science class that ice cream would freeze faster if it was heated first before being put in the freezer. The laughter ended only when a school inspector tried the experiment himself and vindicated him.

Explore further: What is Nothing?

More information: Mpemba effect - Wiki article;
James D. Brownridge web page;
Mpemba Effect scientific paper, March 2010, by James D. Brownridge;
via Newscientist

Related Stories

'Cold' Mars Could Have Harbored Liquid Water

Jun 01, 2009

(PhysOrg.com) -- A new NASA study provides further evidence that Martian minerals dissolved in water could have kept that water from freezing, even on a cold, early Mars.

Scientists isolate new antifreeze molecule in Alaska beetle

Dec 14, 2009

Scientists have identified a novel antifreeze molecule in a freeze-tolerant Alaska beetle able to survive temperatures below minus 100 degrees Fahrenheit. Unlike all previously described biological antifreezes that contain ...

Ice Cold: Cooler Than Being Cool

Nov 24, 2009

(PhysOrg.com) -- Water expands when it freezes. Anyone who has ever left a can of soda or bottle of water in the freezer too long has witnessed this first hand. So how do plants and animals survive severe ...

Scientists Observe Liquid Water Below Freezing

Jun 24, 2009

(PhysOrg.com) -- Below 0 °C, water turns to ice. But beyond that, or below about -75 °C, the ice may turn back into liquid water. While scientists have previously predicted this phase transition with computer ...

Scientists: Less ice on Great Lakes during winter

Mar 23, 2009

(AP) -- Ice cover on the Great Lakes has declined more than 30 percent since the 1970s, leaving the world's largest system of freshwater lakes open to evaporation and lower water levels, according to scientists associated ...

Recommended for you

What is Nothing?

Aug 22, 2014

Is there any place in the Universe where there's truly nothing? Consider the gaps between stars and galaxies? Or the gaps between atoms? What are the properties of nothing?

On the hunt for dark matter

Aug 22, 2014

New University of Adelaide Future Fellow Dr Martin White is starting a research project that has the potential to redirect the experiments of thousands of physicists around the world who are trying to identify the nature ...

Water window imaging opportunity

Aug 21, 2014

Ever heard of the water window? It consists of radiations in the 3.3 to 4.4 nanometre range, which are not absorbed by the water in biological tissues. New theoretical findings show that it is possible to ...

User comments : 36

Adjust slider to filter visible comments by rank

Display comments: newest first

broglia
1 / 5 (3) Mar 26, 2010
..Brownridge said the hot water cools faster because of the bigger difference in temperature between the water and the freezer...
Frankly, I didn't understand this explanation at all.

It would mean, for faster cooling of water we should always heat it before placing it into refrigerator, which is an aparent nonsense, until refrigerator isn't covered by thick layer of ice, which serves as a heat insulator (I presume, Brownridge avoided this trivial mistake in his experiments).

In addition, when the water reaches temperature of cold water (which would require ten minutes, for example), then the speed of further colling should remain the same, because of the same temperature of both samples - just because of the above rule. So we still have ten minutes difference.
BozoQed
4.5 / 5 (2) Mar 26, 2010
Could it be that hot water has more convection in it resulting in a better distribution of the cold...

(sorry if this is BS I am not a physicist)
Skeptic_Heretic
3.7 / 5 (3) Mar 26, 2010
The reason why it makes sense is because of energy content and vibration.

What's the fastest way to cool your coffee in the morning? Stir it.

The energetic particles in the heated water are smashing into each other and transferring energy chaotically, however, at the surface the most energetic are recieving large amounts of energy from this contact and leaving the system subtracting a far greater amount of energy through evaporation than radiative cooling allows for. In the case of covered containers, this allows for more molecules to contact the radiative surfaces of the container and as such also ensure higher amounts of radiative cooling as a poster above me suggested.

Basically, the triple point of water shows us some very "cool" behavior.
bsummey
5 / 5 (1) Mar 26, 2010
This really is scraping the bottom of the barrel. Interesting topic, but terrible explanation. There is more information in one paragraph of skeptic heretics response.
broglia
4 / 5 (1) Mar 26, 2010
IMO the Mpemba paradox is rather connected with repolymerization of water clusters after cooling, which requires some time. Cold water is composed of clusters, which are more prone to crystalization. Such hypothesis could be tested by measurements of dynamic viscosity at nanometer scales, for example by passing of water through porous membranes. Due the limited scope if hydrogen bonds this effect doesn't apply at larger distance scale.
broglia
4 / 5 (1) Mar 26, 2010
There is more information in one paragraph of skeptic heretics response.
Radiative cooling would faster cooling only up to temperatures of cold water - after then the speed of colling will be the same. The explanation of BozoQed seems to be more relevant for me (the more convection currents in a sample, the less super cooling should occur).
Skeptic_Heretic
3 / 5 (1) Mar 26, 2010
Certainly not broglia/alizee. You have a higher rate of conduction and convection due to natural equilibrium, not due to spontaneous organization. Water is not sentient.

The reason why porous membrane organization occurs is because the membrane is responsible for putting the molecules in a row.
antialias
5 / 5 (1) Mar 26, 2010
But shouldn't convection also slow down to the level of the cold water as the hot water cools (particularly if the cooling period is over the order of minutes rather than seconds)

I didn't fiond this "hot water cools faster because of the bigger difference in temperature between the water and the freezer... " particularly convincing. It runs up against the Achilles paradox.
Skeptic_Heretic
2 / 5 (1) Mar 26, 2010
antialias,

Actually as convection increases the energy difference between the ambient water molecule environemnt and the energetically charged molecules increases at a greater rate due to increasing heterogenity, which is the focus of the article above. This is why the article appears to be useless, the explanation is quite poor.
jamey
5 / 5 (1) Mar 26, 2010
The problem that strikes me with this story is that the graph includes the time to take the water samples from their initial room temperature equilibrium to their "start temperatures". That kind of makes me wonder.
Just_some_guy
5 / 5 (3) Mar 26, 2010
People, please! The article is quite clear. I suggest you read it again. The fact that the warm water cools faster is beside the point. The effect is caused by the difference in freezing point between the samples to begin with and this in turn depends on the level and type of impurities in the samples. If you switch the samples around and heat the other one instead, then it would not freeze first.
vanderMerwe
4.7 / 5 (3) Mar 26, 2010
(two parts)

Just a note: Mpemba, a Tanzanian HS student, discovered the effect while making ice cream, not freezing water.

You don't need physics as such to solve this, just a bit of common sense about how refrigerators work. If you put something cool in a fridge, it is much less likely to trigger the thermostat that turns on the compressor than if you put something hot in the same fridge. Thermostats employ a "dead band", that is, the compressor that cools the fridge will come on at one temperature and not shut off till the temperature reaches a point considerably below the temperature at which the compressor was started. That's a deadband, the temperature difference between when the thermostat tells the compressor to start chilling down the fridge and the temperature at which it tells the compressor that things are cool enough.

vanderMerwe
4.8 / 5 (4) Mar 26, 2010
(last part)

The bigger the deadband the less often the compressor is turned on. That extends compressor life. If a fridge thermostat has a big deadband, putting something hot in it will be frozen faster than something cold simply because the hot sample triggers the thermostat and brings more cooling capacity to the sample. You won't see the effect if your fridge has a more modest dead band.

As an old environmental controls professor, I am often astonished at the blinders that PhD physicists too often put on themselves. :-D
physpuppy
3.7 / 5 (3) Mar 26, 2010
Interesting article, good discussion here from everyone.

An interesting article about the subject:

http://www.newton...9631.htm

Does high quality distilled/deionized water show this effect?
One possibility not investigated here is the possibility of dissolved gases making a difference - heating water tends to "degas" or remove oxygen and CO2 from the water. Other methods of degassing would be to bubble other gases through the water - I would imagine helium would be most effective for this type of test - and yet another widely used technique is to apply a vacuum to the water. It would be interesting to see if water treated using any of these methods would have different rates of freezing.

Put all samples in the freezer at once, would probably solve @vanderMerwe's suggestion that the heated water causes the freezer's compressor to start and increase cooling capacity.

@vanderMerwe, the word for what you are describing - "deadband" is "hysteresis".
vanderMerwe
5 / 5 (1) Mar 26, 2010
physpuppy: No, it's a deadband. Hysteresis is something else entirely. Vaguely similar but substantially different.

http://en.wikiped...Deadband

http://www.energy...0602.pdf

As well, if you put all your samples in the fridge at once you'll never be able to separate out the samples' interaction with each other.

You may call yourself physpuppy, but you need to review your experimental design notes. :-/

Skeptic_Heretic
3.5 / 5 (2) Mar 26, 2010
(two parts)
You don't need physics as such to solve this, just a bit of common sense about how refrigerators work. If you put something cool in a fridge, it is much less likely to trigger the thermostat that turns on the compressor than if you put something hot in the same fridge. Thermostats employ a "dead band", that is, the compressor that cools the fridge will come on at one temperature and not shut off till the temperature reaches a point considerably below the temperature at which the compressor was started.

Are you assuming the scientists performing the experiment are that stupid? The samples went into the same fridge. It's called a controlled environment.

Likewise with the statement about nucleation. Both samples were exact other than temperature.

Phys, you do bring up a good point about dissolved gasses. I can't say they accounted for that.
vanderMerwe
not rated yet Mar 26, 2010
"Are you assuming the scientists performing the experiment are that stupid?"

I've seen it happen more than once.

"The samples went into the same fridge. It's called a controlled environment."

If you really think that a fridge is a controlled environment and that samples don't interact, I have a bridge I can get you a good price on.

Skeptic_Heretic
3.5 / 5 (2) Mar 26, 2010
If you really think that a fridge is a controlled environment and that samples don't interact, I have a bridge I can get you a good price on.

Ok, same cooling unit.

Go read the paper before you criticize the controls that you are completely unaware of.
vanderMerwe
3.7 / 5 (3) Mar 26, 2010
I'm going to explain this one more time in little words and let that be an end of it. Fridges, even lab fridges, are NOT A FIXED environment. They interact dynamically with what you put in them. If you have a very big fridge, like a walk-in cold room with an air lock, that interaction is going to be small if your sample is relatively small. If it's your average lab fridge, though, that interaction can be quite substantial. An average sized lab fridge is going to have a very different response depending on whether you put a hot or cool sample in it. If you can't get your head around that. Fail.
jgelt
not rated yet Mar 27, 2010
It really doesn't matter if you heated the water up or not. Nucleation prevents the supercooling.
So the cooler can get 80 degree water down to zero faster than it can get 0 degree water down to -15, according to the chart.
seneca
3 / 5 (2) Mar 27, 2010
We should say clearly, there is absolutely no reason for hot water to cool faster in carefully designed thermostat device. For example, we can place hot water in flat layer to surface of silver or copper dish and cool it nearly immediately in such arrangement. In future experiments the effects of various cooling speed should be separated from effects of various nucleation speed bellow zero temperature - which is what remains crucial for understanding of Mpemba effect.
eric96
1 / 5 (1) Mar 27, 2010
@vanderMerwe

That the fridge is triggered faster is irrelevant, but the extended length it will run for for warm/hot water is significant. So it goes like this: extra cooling from fridge + warm water convects faster + (no coat of isolating frost is created unlike cold water) = faster freezing
rocketmonkey
5 / 5 (1) Mar 27, 2010
Forget hot water freezing slightly faster than cold water - sometimes. The biggest news in this story is that Aristotle had a freezer!
eric96
3 / 5 (2) Mar 27, 2010
@vanderMerwe

That the fridge is triggered faster is irrelevant, but the extended length it will run for for warm/hot water is significant. So it goes like this: extra cooling from fridge + warm water convects faster + (no coat of isolating frost is created unlike cold water) = faster freezing. One must think of ice in a similar way to a melting point. With cold water...its a short race track....with a turtle as the runner. With warm/hot water, its a much longer race track, but with a cougar as the runner. Basically, speed is more relevant than distance. Hot water wins. The convection effect of hot water would be nullified if both placed in copper trays.
jared_mackay
not rated yet Mar 27, 2010
I think some of the confusion in the article comes from this sentance:

Brownridge said the hot water cools faster because of the bigger difference in temperature between the water and the freezer, and this helps it reach its freezing point before the cold water reaches its natural freezing point, which is at least 5°C lower.

He seems to have mixed up two of the many experiments in the paper. What the paper actually says is that people often observe what they think is the Mpemba effect because the hot water interacts with freezer in some way, such as melting freezer frost, which then refreezes into ice which has a higher thermal conductivity than frost. From the conclusion (in BOLD!)

Hot water will freeze before
cooler water only when the cooler
water supercools, and then, only if the
nucleation temperature of the cooler
water is several degrees lower than
that of the hot water. ***Heating water
may lower, raise or not change the
spontaneous freezing temperature.***
jared_mackay
not rated yet Mar 27, 2010
This article is very poorly worded. The author doesn't seem to have understood the research very well, possibly he mixed up two of the many experiments in the paper. What the paper actually says is that people often observe what they think is the Mpemba effect because the hot water interacts with freezer in some way, such as melting freezer frost, which then refreezes into ice which has a higher thermal conductivity than frost.

For two samples of pure water, the temperature difference, even up to 100'C has less importance than the nucleation ("freezing") temperature of the water sample.

From the conclusion (in BOLD!)

Hot water will freeze before
cooler water only when the cooler
water supercools, and then, only if the
nucleation temperature of the cooler
water is several degrees lower than
that of the hot water. ***Heating water
may lower, raise or not change the
spontaneous freezing temperature.***
jared_mackay
not rated yet Mar 27, 2010
This article is very poorly worded. The author doesn't seem to have understood the research very well, possibly he mixed up two of the many experiments in the paper. What the paper actually says is that people often observe what they think is the Mpemba effect because the hot water interacts with freezer in some way, such as melting freezer frost, which then refreezes into ice which has a higher thermal conductivity than frost.

For two samples of pure water, the temperature difference, even up to 100'C has less importance than the nucleation ("freezing") temperature of the water sample.

From the conclusion (in BOLD!)

Hot water will freeze before
cooler water only when the cooler
water supercools, and then, only if the
nucleation temperature of the cooler
water is several degrees lower than
that of the hot water. ***Heating water
may lower, raise or not change the
spontaneous freezing temperature.***
Salander
5 / 5 (2) Mar 27, 2010
It's been four decades since the Mpemba incident, but as I recall it happened something like this.
Mpemba was a very bright student who observed that hot ice cream froze faster than cold. When he reported it to his science class, the students laughed at him. His instructor told him it was impossible. Yet Mpemba persisted because he had seen it himself. He explained his observations to a visiting British physicist who also told him it was impossible. Again Mpemba insisted. His instructor told him he should be ashamed for contradicting so high an authority. Finally, the visiting physicist, out of curiosity, invited Mpemba to demonstrate. The physicist was amazed but convinced when the hot ice cream froze faster, and returned home to write the paper which made Mpemba famous.

He deserves his fame because he displayed the true qualities of a scientist: he observed reality and refused to racant his observations when authority pressured him. Galileo would be proud.
Bob_Kob
not rated yet Mar 28, 2010
So yeah I put boiling water in ice cube tray and cool water in another. The cool water froze first.

Whats the exact conditions that this effect takes place?
seneca
not rated yet Mar 28, 2010
Whats the exact conditions that this effect takes place?
Many, in fact. You should consider, refrigerator cools more intensively, when you place hot watter into it. You should consider, hot water evaporates faster, so that there remains smaller amount of water to cool at the end. You should consider, hot water melts the ice covering refrigerator and/or cooling vessel, thus providing better cooling contact. You should consider, hot water has a lower viscosity and larger gradient of temperatures provides better conditions for convective cooling. All samples should be outgassed and free of impurities, which could crystallize during cooling, thus serving as an ice nuclei. Just after all these effects are eliminated, you can start to investigate Mpemba effect in reliable way.
HoboWhisperer
5 / 5 (1) Mar 28, 2010
@vanderMerwe
I think you are definitely onto something with your deadband hypothesis - I also have quite a bit of experience with industrial temperature control systems (PID and on-off). Control systems are never linear, different disturbances of different magnitudes (different temperature samples) WILL result in different dynamic behavior.

Of course that is not to say this is the only effect at play.
ubavontuba
1 / 5 (1) Mar 28, 2010
Hmm... It seems what everyone is missing is the convection in the refrigerator caused by the heated water element. That's to say, the heat radiating off of the heated container will cause convection to occur in the fridge's air space. The heated air rises away from the container, and the cooler air falls and surrounds the container. This convection would continue for some time, even after the container cooled to a lower temperature than the low temperature, control container.

In short, the hot container is exposed to more cold air.

It's the opposite effect of a convection oven. Everyone knows foods cook faster in a convection oven, as opposed to a standard (static) oven.

I also like the deadband hypothesis.
ubavontuba
1.5 / 5 (2) Mar 28, 2010
rocketmonkey:
Forget hot water freezing slightly faster than cold water - sometimes. The biggest news in this story is that Aristotle had a freezer!
An excellent observation!
seneca
3.3 / 5 (3) Mar 29, 2010
As Aristotle says in his book on drunkenness, those who have drunk the barley liquor called beer fall on their backs; he actually said: "The liquor made from barley called beer has a certain peculiarity; people who are intoxicated by other liquors fall in all sorts of directions to the left, to the right, on their faces, on their backs; only those who are intoxicated with beer always fall backwards and lie on their backs..."

So, if Aristotle had a beer while lying on back in his couch - he would need some freezer to complete consumer experience, too...
physpuppy
2 / 5 (1) Mar 30, 2010
@vanderMerwe - I looked at your references regarding the difference between hysteresis and deadband, and there are a few things that I don't understand.

One article states:
"Thermostats exhibit hysteresis. [assumption here: not deadband]" I assume that lab fridges are controlled by a thermostat. To clear it up, the other reference states:
"A deadband thermostat is essentially two independent thermostats in one housing. One thermostat controls heating and the other controls cooling."

Unless the fridge is self defrosting - as most consumer and some of the non-critical lab fridges are, but not most lab fridges - it would not need such a thermostat.

With Deadband, there is a range in which no change is made - I would assume that temperature varies during that time..

Lab fridges are made to control temperature fairly precisely. [in my experience, not accurately] In consumer fridges temperature can vary quite a bit (and as such I can understand your comment about deadband)
femy
5 / 5 (1) Apr 14, 2010
i think the hot water molecules have more average kinetic energy than cold water so its molecules easily come to surface and evaporate .on the other side cold water have more density than hot water and there molecules will have less kinetic energy and they will face a lot of resistance from the neighbouring atoms and will transfer some of energy to the neighbouring atoms and since it have already less kinetic energy it will become least and hardly reach reach to the surface and will take more time.since freezing is the result from evaporation of high energy molecules therfore cool water will take more time in freezing......