Would a glass of water in space freeze or boil?

This was a Cecil classic that I stumbled upon today, and I think I disagree with Cecil on this one.

When I was in high school, the teacher performed an experiment for us - into a vacuum bell jar, he placed a glass of tap water.

As the vacuum was pulled, the water began to boil and freeze simulataneously. According to the teacher, the pressure was lowered enough that the crytal structure could form, even at room temperature.

The only difference I can see from deep space is ambient temperature - in the classroom, there would still be radiative coupling to room temperature, as opposed to the 2.5 Kelvin of deep space.

here is a phase diagram of water. If we assume that the pressure of space is essentially 0Pa then a temperature of around 200K would be required to vaporise the water. In deep space the water would therefore be frozen.

Here’s the link to the article in question. (For future reference, it’s considered de rigueur to include a link to the article.) I’m not certain what disagreement you have with Cecil’s conclusion. From the article:

One factor to consider as well is whether the water in question is being evacuated in sunlight or shade; the surface temperature of the Sun is about 5800 Kelvin, so depending on your proximity to the Sun and other radiant bodies your water may be absorbing more radiation than it radiates, which might make it at least transiently liquid vapor.


I think tomii’s disagreement is the part about the water boiling, followed by the vapor passing to the solid state. Instead, [s]he observed “the water began to boil and freeze simulataneously”. I believe tomii’s rembered observation is correct. Looking at Fast ‘n’ Bulbous’s link to the phase diagram, the liquid water is unstable, since it’s in the vapor region (at, say, 280 K and 1 Pa). As the water boils, it takes away heat form the remaining water, so the temparature of the water drops, eventually getting low enough that some of it freezes directly, rather than going through a vapor phase.

I put some numbers to this. The heat of vaporization is 539 cal/g at STP, but a physics book I have says about 40 of those go to the work of expanding into the atmosphere, so I’ll use 500 cal/g. The heat of melting is 80 cal/g from the same book, and water changes temperature at 1 degree per cal per gram. I’ll assume those STP numbers are still valid because I don’t know any better.

I’ll start with 100 grams of water at 280 K, and use 200 K as the temperature of the vapor/solid phase change. Vaporising 14 grams takes 14*500 cal, removing enough energy to lower the temperature of the remaining 86 grams of water to about 200 K. Vaporising another 12 grams takes about enough energy to freeze the remaining 74 grams of water. So about 3/4 of the water ends up frozen, and only 1/4 ends up as vapor.

Careful about inferring actual dynamic behavior, particularly at extreme conditions, from the phase diagram; the phase diagram represents a quasistatic equilibrium that assumes local constant temperature and pressure. Back when I had to (attempt to) model the dynamics of fuel combusion in an internal combustion engine, using the equations of state along with mechanical losses gave me limits for initial and final conditions but told me nothing about the completeness or the efficacy of combusion; in order to optimize the dimensional configuration of the cylinder and parameters for injection of the fuel-air mixture I actually had to model the dynamic effects of the local nonequilibrium state, which is an extremely complex task even for modern computational fluids simulation codes. And talking about the water “taking away heat from the remaining water”, as if it as is a mechanically-continuous medium is somewhat misleading; a fluid exposed to sudden decompression sufficient to vaporize it will act more or less like an ideal gas.

As a practical matter I can’t see that there’s much difference between Cecil’s prediction (“The vapor then passes immediately into the solid state…and you end up with a cloud of very fine crystals of frozen urine,”) and tomii’s observation, save that in the scenerio Cecil postulates, i.e. an open vacuum microgravity environment, the vapor can expand indefinitely, whereas tomii’s case has the water in a not-quite-adiabatic closed cell.


The difference is that Cecil’s description has the water first vaporizing, then freezing. What my calculations show is that there isn’t nearly enough energy in the system to break all the bonds. It simply can’t all vaporize first.

I’d be interested in hearing some plausible reason how the water could possibly all vaporize; I’m not seeing it. I realize I’m making a back-of-the-envelope estimate here, but the numbers I started with would have to change a lot to allow even half the water to vaporize. Because of the approximations I made, I end up with half the vapor at 280 K, and the other half at 200 K, when we’d expect it to be a continuous distribution over that range, but the average should be close, and doesn’t have that much of an effect anyway. The vapor that boils away, and any ice or water it pushes away, would have some kinetic energy, but that would mean there’s even less energy available for breaking bonds. Where’s the energy for breaking all the bonds coming from?

Comets are balls of mud and ice. The ice in comets exist in space over long periods of time, but the comet tail indicates the sun can slowly boil away surface ice. That’s how I envision the the glass of water: it would turn to ice, but the sun can slowly boil away the surface which disperses away.

First of all, I would caution against making calculations based upon the phase diagram, which represents the condition of water in a quasi-equilibrium state. It is possible for water to exist in other phases in a dynamic or metastable condition, i.e. superheated and supercooled liquid, although the slightest impetus will cause it to fall into its normal phase for those conditions.

Second, and this is my error (or at least failure to be sufficiently pedantic), but I was using vapor in the colloquial sense which includes aerosols, rather than the more specific term gas. What I would envision is that the glass of water would aerosolize in a vacuum, creating tiny droplets of water (the size of which is dependant upon the balance between surface tension and internal pressure), which would then freeze as they lose heat via evaporative cooling and radiation, then condensing onto crystalline flakes akin to snowflakes. I would WAG that only a modest amount of the initial liquid would permenantly break the hydrogen bonds and go to a gaseous state. I would be surprised if anything more than a skim of water adhering to the surface of the glass remained in a solid continuum (i.e. a block of ice).


Link (PDF warning) to a discription of a refrigerator based on water using a vacuum maintained by zeolite sorbant.

tomii, I also (tentatively) disagree with Cecil on this one in so far as I don’t think there’s enough energy in the water for it to vapourize completely, and in this I’m in agreement with ZenBeam, who correctly points out that as the water boils it loses heat to the extent that it starts to freeze.

Maybe the water first starts to boil, and then starts to freeze. Are you sure you saw the water start to freeze as soon as it started to boil?

I confidently disagree with the idea that lowered pressure leads directly to freezing of water even at room temperature. Increased pressure perhaps could do it.

Do you know if the vacuum was “pulled” instantly or did the pressure drop slowly as air was pumped out of the chamber. Please clarify what you mean by “pulled”.

Having looked at the phase diagram of water kindly provided by Fast ‘n’ Bulbous I see that by increasing the pressure enough the water can indeed be solidified at room temperature, or, it seems, at hundreds of degrees Celsius or maybe more than that.

Of course, it’s impossible to pull a vacuum “instantaneously” on earth. In the lab, it was a small bell jar hooked to a 2 cylinder pump - I imagine the pressure dropped pretty quickly. It’s been over 20 years ago, so I honestly don’t remember, but I imagine that it probably did start boiling before it froze.

Yes, you’re right, but it could be made much quicker by pumping air out of a large container and then suddenly connecting the water to that by means of a quick opening large valve. In which case I think you’d see something like explosive boiling of some of the water, followed almost instantly by freezing of some of the water.

Explosive boiling of water can occur when a cup of water is heated up in a microwave oven and gets superheated. The water is well over 100 degrees but it’s kind of not got around to boiling. It can be started by adding a teabag, or just a tea spoon and once it starts, watch out :slight_smile: I’ve seen this. It’s dangerous. Apparently this superheating of water is the main cause of harm by microwave ovens.