Why does a hot or warm ice cube tray freeze quicker than one at 33*F?

[capnahab]

I have the answer, it is because the refrigerator is not in freeizng mode when the cold ice cube tray is placed inside. When the warm tray of water is placed inside then the temperature lowers considerably and the thermostat switches the freezer into freezer mode. A common complication I encounter with a ammonia blast freezer on a regular basis. Not being a mechanical engineer, is there momentem with heat?

 

[Gokul43201]

I imagine this experiment has been performed in conditions other than commercial refrigerators with thermostat controlled freezers!

If you want to look it up further, it's called the Mpemba Effect.

 

[capnahab]

OK, you all give me the reason that hot water freezes faster than cold water? And by observation!

You all need to study the primice of mechaical engineering! If it hurts, don't do it.

 

[russ_watters]

Unless there is an inequality to the experimental setup, such as a reduced mass/volume of warm water due to evaporation, this can't be true. The very fact that the hot water should pass through the temperature the cold water started at, at which time it will be in an identical condition makes it so.

We've had this question before - perhaps I'll do a scientific test with a data-logging thermometer next week (have to get it from work)...

 

[Gokul43201]

The very fact that the hot water should pass through the temperature the cold water started at, at which time it will be in an identical condition makes it so.

But it does not - at least in the sense that it's not sufficient to just look at the mean temperature of the water. Fast cooling is not an equilibrium process. The water that started out warm need not (and if undisturbed, will not) ever have an identical profile to the water that started cold.

 

[TVP45]

Unless there is an inequality to the experimental setup, such as a reduced mass/volume of warm water due to evaporation, this can't be true. The very fact that the hot water should pass through the temperature the cold water started at, at which time it will be in an identical condition makes it so.

We've had this question before - perhaps I'll do a scientific test with a data-logging thermometer next week (have to get it from work)...

About 15 years ago, this experiment was performed in a large, well-controlled environmental chamber at one of the corporate R&D centers here. We used -20C and shielded the trays from convection. Humidity was not controlled. Nobody was able to state, based on observation, which froze first. Thermocouples gave no clear differentiation as their reading very much depended on their location in the trays. The two ladies assisting me swore they froze first, but that was felt to be subjective. Two repeats gave the same lack of clear differentiation. However, containers shaped similar to buckets seemed to show the cold freezing first, at least on the surface. We thought about stirring, but that seemed at odds with the so-called Mpemba effect. If you do this, please share the results - I am greatly suspicious of this effect.

 

[russ_watters]

But it does not - at least in the sense that it's not sufficient to just look at the mean temperature of the water. Fast cooling is not an equilibrium process. The water that started out warm need not (and if undisturbed, will not) ever have an identical profile to the water that started cold.

Ok, true, there will be a gradient. But the cold water will also form a gradient as well, so the warm water will have to at some time act pretty darn similar to the cold water.

 

[ZapperZ]

Here's a bunch of references to this effect:

http://arxiv.org/abs/0704.1381
http://arxiv.org/abs/physics/0604224
http://arxiv.org/abs/physics/0512262

Zz.

 

[belliott4488]

About 15 years ago, this experiment was performed in a large, well-controlled environmental chamber at one of the corporate R&D centers here. ... The two ladies assisting me swore they froze first, but that was felt to be subjective.

Dang! The ladies froze?? You must have a hard time keeping assistants in your lab!

 

[TVP45]

Here's a bunch of references to this effect:

http://arxiv.org/abs/0704.1381
http://arxiv.org/abs/physics/0604224
http://arxiv.org/abs/physics/0512262

Zz.

Zz,

Thanks for the references. My concern is that the problem (or effect) does not seem to be clearly stated. Is the effect about adiabatic cooling through evaporation or about supercooling in closed pipes? Is this distilled water or hard tap water?

Before delving into explanations for why this occurs, I'd like to see some repeatable evidence that it really does occur. The original Mpemba event leaves me cold (1st bad pun) for these reasons:
(1) He compared boiled milk to raw milk. Boiling changes milk and the more one boils it, the greater the change. In Africa, milk is commonly boiled for extanded times and, so, that needs to be clarified (2nd bad pun).
(2) The raw milk guy was clearly "short-cutting" the lab. Did he also omit the sugar and the vanilla (I understand they were making ice cream)?
(3) How did he determine freezing? In the absence of a sensor, freezing of an opaque liquid with suspended solids may be difficult to determine.
(4) Does the experiment ring true? One doesn't normally attempt to make ice cream by freezing milk. Try freezing Peachy Paterno sometime; it just doesn't crystallize (3rd really bad pun).
(5) How did he extrapolate from milk to water?

Again, I hope I remain open to the possibility of there being a Mpemba Effect, but I would want to see better defined control of variables.

 

[ZapperZ]

Did you actually read the first paper? They did the experiment rather carefully.

Zz.

 

[TVP45]

Zz,

Yes, I did. But, and I admit I'm being skeptical, I don't believe I could answer yes or no about the Mpemba Effect based on this paper. Has it been peer-reviewed?

I guess I would say that I can believe that, under certain circumstances, water that was recently hot will freeze faster than water than was not. But, the Mpemba claim, as I understand it, is that a tray of hot water freezes faster than a tray of cold water. Is there evidence for this?

 

[TVP45]

Dang! The ladies froze?? You must have a hard time keeping assistants in your lab!

Do you not have a hot tub in your lab?

 

[ZapperZ]

Zz,

Yes, I did. But, and I admit I'm being skeptical, I don't believe I could answer yes or no about the Mpemba Effect based on this paper. Has it been peer-reviewed?

S. Esposito et al., Physica A: Statistical Mechanics and its Applications, v.387, p.757 (2008).

Zz.

 

[Gokul43201]

Yes, there are several peer-reviewed papers on Mpemba.

For a synthesis, see JB's webpage on it, and references therein:

http://math.ucr.edu/home/baez/physics/General/hot_water.html

 

[TVP45]

Yes, I was familiar with the paper you linked. And, I do accept that hot water sometimes freezes faster than cold water; for example, I have seen the effects of spraying hot water on a car when the outside temperature is a stable -2C or so.

My question is whether these papers address the common extension of the Mpemba Effect which says "an ice tray of hot water freezes faster than one of cold water".

And, frankly, I question the original Mpemba Effect. I just repeated that as best I could determine and my cold milk froze quite a bit faster than the boiled milk. And, neither result looked anything like ice cream.

I sometimes make ice cream and, on several occasions, have used unchilled mix because I was rushing. I saw that it took a lot more salt and ice to freeze (and, of course, more time to churn), but the result was noticeably lighter and tastier.

Anyway, I welcome any input from others who might have tried the original experiment.

 

[russ_watters]

Here's my experiment.

First picture is of the experimental setup. An effort was made to ensure consistent placement of the probes. The cups and probes were set up, then 9 oz water from different sources was poured into a graduated shot-mixer, then into each cup. Then the whole setup was placed in the freezer. You can't see the starting temps from the graph (I zoomed into the freezing area), but they were about 170F, 106F, 68F, and 48F. There was some effect of using the same mixer and the room-temp start of the cups, ie the 170F water came straight off the stove, boiling, the 106F water came from the tap probably at about 125F, and the 48F water came from the fridge, probably at around 35F. I doubt that this had an impact on the test.

My water is unusually hard.

Some of the temperature probes were off by as much as 1.5F, and I performed a second experiment with one cup and all four probes rubber-banded together to normalize the data. I normalized to 32F, but I don't know if the real freezing temp might have been lower due to the dissolved minerals. This isn't critical though, only the timing is critical.

As you can see from the graph, the test failed to find the effect. Assuming freezing starts the instant the subcooling breaks, they started to freeze, in ascending order of starting temp, at 32:50, 49:40, 1:15:40, and 1:30:10. The intervals follow the temperature slope almost exactly.

Spikes that happen simultaneously are likely due to the compressor cycling. I did not open the door during the test.

I have no explanation for the temporary flattening of each cooling curve at 37F. It appears the first paper ZZ linked also showed this "novel transition".

The first paper ZZ linked suggested the possibility that the initially hot water might undergo less subcooling. My results don't show this. However, given just how much supercooling they show, the impurities in my water may make it tougher to supercool. So they may be right that the supercooling is where the effect comes from.

Yes, I have too much time on my hands.

 

[russ_watters]

Here's the data normalized to line up the start of the first phase transition. The water that started cold seems to take less time to freeze, but the others are statistically identical. I don't have an explanation for this.

 

[Integral]

Good work Russ,
Looks to me like there is nothing surprising or unexpected in your data. Clearly the cold water (Fridge line) froze first. I am putting this in the urban legend category.

 

[russ_watters]

One thing about this issue that makes it murky is there doesn't seem to be a consensus on just what the effect is. Is it the time to the start of freezing? The end of freezing? The duration of the freezing? Subcooling and disruption of subcooling makes for a considerable variability in when the freezing begins and can also change the time it takes to freeze. That makes the effect - if it exists - very tough to quantify and implies it may just be a matter of confirmation bias in random data.

 

[TVP45]

One thing about this issue that makes it murky is there doesn't seem to be a consensus on just what the effect is. Is it the time to the start of freezing? The end of freezing? The duration of the freezing? Subcooling and disruption of subcooling makes for a considerable variability in when the freezing begins and can also change the time it takes to freeze. That makes the effect - if it exists - very tough to quantify and implies it may just be a matter of confirmation bias in random data.

Russ,
Thanks for sharing your data. And, I agree with your questions about exactly what the effect is. I suspect this'll still be a recurring question 50 years from now?