# Why does water ($\mathrm{H_2O}$) only have two distinct fluid phases?

Water (and other substances) can exist in many distinct solid phases (with different crystallic micro-structure), but only in two fluid phases - liquid and gaseous, in which the molecules are oriented randomly (they is no long range order). Is there an explanation in the molecular theory, why there are just two "disordered" phases? Why isn't there just one? Or more than two?

• Arguable the "gas" and "liquid" phases are not even actually distinct in a global view, not withstanding that they are quite distinct for practical purposes at the pressures and temperature where we encounter them. See also "critical point" Nov 28, 2013 at 5:07
• What do you mean by "global view"? At given temperature and pressure, gaseous and liquid part of system are distinct phases, they have distinct density and are spatially separated. The phases can be recognized unambiguously - there is a phase interface where density changes discontinuously, similarly to solid - gas interface. Nov 28, 2013 at 23:08
• There is a path in P-T space to take a clearly "liquid" sample to a clearly "gaseous" sample without a phase change. You carry it up in pressure and temperature until you are able to pass around the end of liquid-gas boundary and back down again. That is what I mean by "global view". And your argument is what I mean by them being "quite distinct for practical purposes". Both of these statements are true. Nov 28, 2013 at 23:16
• All right, but my question is about situation when both liquid and gaseous phases co-exist and are distinct, that is below critical point. In this situation, why there can be at most $two$ (and not three or more) distinct fluid phases in the first place ? Nov 29, 2013 at 3:28
• Helium has three fluid phases, gas, liquid, and superfluid Jan 12, 2018 at 19:48

The most immediate answer would seem to be that a great variety of different crystal phases can exist because their long-range order makes it possible to classify them based on the different symmetries of their lattice structure. Since the liquid (or amorphous solid) phase only has short-range order and the gaseous phase doesn't even have that, it seems impossible for different fluid phases to exist.

However...

It turns out that it is possible for an amorphous substance (glass or liquid) to exist in different stable phases. This phenomenon, which is the amorphous counterpart of the polymorphism of crystal, is known as polyamorphism.

Quoting from Wikipedia:

Even though amorphous materials exhibit no long-range periodic atomic ordering, there is still significant and varied local structure at inter-atomic length scales (see structure of liquids and glasses). Different local structures can produce amorphous phases of the same chemical composition with different physical properties such as density.

One example is the liquid-liquid transition exhibited by some model systems, in which a transition from a low density to an high density liquid state appears.

The presence of a liquid-liquid critical point has been hypothesized to explain some thermodynamic anomalies of liquid water. Unfortunately, it is extremely difficult to reach this critical point experimentally, because the system undergoes spontaneous crystallization.

But it has been found out in numerical studies of some simple model systems of water that a liquid-liquid critical point is indeed present, and two distinct, stable liquid phases appear: a low density and high density liquid.

As far as gas are concerned, the absence of local structure (short-range order) makes it impossible for different phases to exist. The only exception which comes to my mind, if we want to call it a "gas", is the Bose-Einstein condensate, obtained when a dilute gas of bosons is brought to temperature close to absolute zero.

• Oh, cool! This is why it's worth bringing old questions back. Can you elaborate on what aspects, beyond only density, distinguish the two different liquid phases? In addition, are the 'simple model systems' theoretical constructs or actual experiments? (As far as I can tell, your link gives only numerical evidence, but it's not easy at all for a general reader to disentangle that.) Jun 30, 2016 at 1:17
• @EmilioPisanty Yes, that was a numerical study with a simplified model. I am not sure if the transition was ever observed experimentally...For water, I am sure it was never observed. But I should check out the literature. As for the first question, structural quantities like the radial distribution function and the structure factor should be different, but again I should check if there is some article with the data :-) Unfortunately I am very busy at the moment, but I will see what I can find as soon as possible. Cheers! Jun 30, 2016 at 14:36
• Cool. As far as I'm concerned this is already worthy of the bounty (unless someone comes and trumps you), but I'll leave it up for the week for the advertising. In the meantime, it's probably worthwhile making that numericalness explicit in the post itself. Jun 30, 2016 at 14:49
• @EmilioPisanty Done! Now it should be more clear. Anyway, I will be glad to add some details as soon as possible! Jun 30, 2016 at 14:53
• @valerio92: Could it be possible that that is just an artifact of an oversimplified model? Aug 8, 2016 at 0:31

Nothing in the laws of thermodynamics forbids multiple liquid phases for a single substance. The only limit is the simultaneous coexistence of at most three phases (at triple points).

Water has a solid-liquid-gas triple point and several soid-solid-liquid and solid-solid-solid triple points; see the phase diagram of water and ice. In addition, although not visible in this diagram (showing only stable phases), it is not true that water has only two fluid phases. Indeed, in the supercooled regime, there are two distinct liquid phases that may coexist under certain conditions. See

P7 Supercooled water has two phases and a second critical point

in http://www1.lsbu.ac.uk/water/phase_anomalies.html, and (one of many publications) http://pubs.acs.org/doi/abs/10.1021/bk-1997-0676.ch018

A recent paper by Holten et al. describes quantitatively, in numerical detail, the thermodynamics of supercooled water, taking into account all experimental data up to 2012.

With solids atoms are mostly locked in place so it makes sense there can be lots of different crystal structures and atomic packings.

For liquids and gases though their defining characteristic is that their atoms are mobile enough to flow and fill a container. You can't both have structure and mobility.

• I suppose one could argue that a plasma is fluid-like but I don't really think plasmas "count". Nov 28, 2013 at 1:01
• Plasmas are definitely fluid-like, though they're electrically conductive so you have to use MHD rather than HD, and definitely a state of matter, even so far as it is actually the most predominant state of matter. Nov 28, 2013 at 2:30
• I do not think this answers my question. I did not demand that the other hypothetical fluid phases had structure; I just wonder why are there just two of them. Comment to Brandon's answer: liquid won't fill any container like gas does. At given temperature and pressure, it has definite density like solid has, but gas has not. Nov 28, 2013 at 23:15
• @JánLalinský liquids do fill a container but the individual molecules don't have enough energy to stop from adhering to each other. I think the primary difference between a liquid and a gas is that a liquid doesn't have enough energy to overcome cohesion. There is a semi-intermediate for some substances called a wet gas: en.wikipedia.org/wiki/Wet_gas Dec 3, 2013 at 6:43
• As far as "having structure and mobility" there are materials called liquid crystals where the constituent molecules are mobile, yet the molecules still have orientational and/or positional order. It is reasonable to ask why water doesn't form such as liquid crystal phase. Jun 29, 2016 at 19:55

There is actually only one disordered phase - from a physicist's perspective, the liquid and the gas are actually the same phase because one can continuously vary the external parameters (temperature and pressure, in this case) to get from the liquid to the gas without passing through any phase transition, because the phase transition line terminates within the phase. (The "liquid-liquid" transition that valerio02 describes is another example of a phase transition within a single phase.) It may seem counterintuitive that you can have a phase transition with the same phase on both sides, but it's true.

Using Landau-Ginzburg theory, one can show that only a single disordered phase can exist in any system, because phases are characterized by the symmetries of the Hamiltonian that they break, and a disordered phase by definition does not break any symmetries.

Things get more complicated when you consider topological phase transitions, which do not have a local symmetry-breaking order parameter and therefore cannot be described by Landau-Ginzburg theory, but that's a more complicated subject that doesn't apply to water.

• But Landau-Ginzburg theory is only a qualitative approximation; it does not even correctly predict the properties at a ctitical point. Jun 30, 2016 at 17:25
• @ArnoldNeumaier That's true (in low dimensions - in sufficiently high dimensions LG theory becomes exact for conventional second-order phase transitions), but it is qualitatively accurate. It gets the critical exponents wrong, but you can usually roughly describe each (non-topological) phase using LG theory. Jun 30, 2016 at 17:47
• In contrast what you write in the answer, "a phase" is not defined as the area of the phase diagram where you can reach without touching any lines. A distinct phases here probably means real observable distinct regions of space under same macroscopic state variables. In other words, I would say 'two distinct phases of liquid' is equivalent of saying: there exists a liquid-liquid phase transition. Jul 4, 2016 at 22:55
• @MikaelKuisma Quantum phases are indeed typically defined to be contiguous regions of parameter space that are not bisected by any phase transitions, because these phases are described by a single order parameter. You are describing the quite different phenomenon of phase separation, where you have a first-order phase transition and you try to to tune a parameter into a region between the two possible phases, so the system breaks up into physical subsystems that lie on opposite sides of the first-order transition. Jul 4, 2016 at 23:12