IF YOU think you lead a high-pressure life, imagine trying to survive in the
Marianas Trench鈥攖he deepest place on Earth, 11 kilometres beneath the
surface of the Pacific Ocean. It is a world of silence and eternal darkness,
where pressures soar to a crushing 1000 atmospheres. A tower of some 10 000
tonnes of seawater bears down on every square metre of ocean floor.
Life is surely arduous under such brutal conditions. But some physicists now
believe that even water itself may have trouble maintaining its composure at
such pressures. They claim that if you could hold water at these pressures, and
keep it liquid while cooling it to temperatures far below its ordinary freezing
point鈥攏o mean feat, to be sure鈥攊t would separate into two utterly
distinct liquids, one about 20 per cent more dense than ordinary water, the
other about 10 per cent less dense. Under these extraordinary conditions, they
say, there is not just one kind of water, but two.
Water is a strange stuff. Unlike most other liquids, it expands when it
freezes鈥攚hich is why icebergs float and frozen pipes burst. Stranger
still, cold water expands even before it freezes, and is more dense at 4 掳C
as a liquid than it is when it turns to ice. Without water鈥檚 weirdness, life on
Earth wouldn鈥檛 be the same. But can there really be two entirely different forms
of the liquid?
Advertisement
The reason why water, of all liquids, might be singled out for this strange
duality is that its molecules have an uncommon stickiness. Molecules in any
liquid attract one another, but H2O molecules do so in a style of their
own, by forming hydrogen bonds鈥攚hich are some ten times weaker than normal
chemical bonds. These hydrogen bonds encourage each water molecule to surround
itself with four others at the corners of a tetrahedron.
This leaves a lot of space between molecules, which is why ice, in which the
tetrahedral arrangement is rigidly enforced, has an open structure. In liquid
water the molecules can drift into some of the gaps, which is why water is
denser than ice. So there is a peculiar tension in the liquid between the
tetrahedral arrangement, which makes for nice, unstrained hydrogen bonds, and
denser arrangements having more efficient molecular packing. This offers the
possibility of a switch in the structure when the temperature and pressure are
changed, as one factor or the other becomes dominant. It鈥檚 long been supposed
that the density maximum at 4 掳C might be an indication of such a gradual
changeover. Perhaps, say physicists, under more extreme conditions the
changeover might be abrupt.
But if two waters exist, how come they haven鈥檛 been seen before? The snag is
that water鈥檚 split personality is predicted to manifest itself only in
supercooled water鈥攚ater that remains liquid even below the freezing point.
Ordinarily, ice forms at such temperatures as molecules link up into a regular
network. But this only happens if, in a process known as 鈥渘ucleation鈥�, something
triggers ice formation鈥攑erhaps a group of water molecules that come
together by chance into the crystalline pattern. This cluster then acts as a
seed for further freezing. But if you remove contaminants like dust particles
which can act as seeds for nucleation, you can defy freezing and obtain
supercooled liquid water.
In practice, there is a limit to supercooling: no one has yet cooled water at
atmospheric pressure to much below 鈭�38 掳C before it freezes. That鈥檚
frustrating, because there are signs that at temperatures just slightly lower
something very interesting happens: water seems to go crazy.
In 1976, physicists Austen Angell at Arizona State University in Tempe and
Robin Speedy, now at the Victoria University of Wellington, in New Zealand,
discovered evidence that supercooled water becomes 鈥渉ypersensitive鈥� somewhat
below the supercooling limit. They estimated that water鈥檚 heat
capacity鈥攖he heat needed to raise its temperature by, say, one
degree鈥攕hould become infinite at 鈭�45 掳C. Other of its properties
also seem to experience some kind of catastrophe there. Why?
Speedy suggested in 1982 that the limit of supercooling of water is
ultimately set by a boundary called the spinodal line, at which this kind of
wild behaviour is anticipated. But in 1992, Gene Stanley, his graduate student
Peter Poole and coworkers at Boston University came up with a different
explanation. Supercooled water, they suggested, might be able to exist as either
a high-density liquid (HDL) or as a low-density liquid (LDL). They also proposed
that the dividing line between these two liquids鈥攖he set of pressures and
temperatures where water would change from one form to the other鈥攎ight end
in a 鈥渃ritical point鈥� at around 鈭�50 掳C and 1000 atmospheres where the
two liquids would become indistinguishable.
Critical point
At a critical point, some of a substance鈥檚 properties (such as its heat
capacity) become infinite. But it doesn鈥檛 happen abruptly. Like Mount Everest
surrounded by the lesser corrugations of the Himalayas, a critical point
distorts the terrain all around, over a large range of temperatures and
pressures. 鈥淣ear a critical point,鈥� says Stanley, 鈥渢hings have to be huge. If
something is going to be infinite at 鈭�50 掳C, it has to be very big at
鈭�20 掳C, or even at room 迟别尘辫别谤补迟耻谤别.鈥� So if there were a liquid-liquid
critical point, even at very high pressure, it might account for the crazy
behaviour of supercooled water at around 鈭�45 掳C and atmospheric
pressure. It could also explain some of the weird properties of water under
ordinary conditions. The density maximum at 4 掳C, for example, might
represent a ghost of the competition at lower temperatures and higher pressures
between the low- and high-density liquids.
It鈥檚 a nice idea. But if the two liquids exist only below 鈭�38 掳C,
that puts them in a no-man鈥檚-land that simply cannot be reached by
experiment鈥攏o one can keep water from freezing at such temperatures. So to
make progress, researchers have abandoned liquid water and turned instead to
ice.
If you can鈥檛 make liquid water below 鈭�38 掳C, you can make the next
best thing: glassy 鈥渁morphous鈥� ice鈥攍iquid water cooled so much that its
molecules stop moving. Just as silica can exist as ordered quartz crystals and
as a disordered, amorphous glass like that in your windows, so too can ice. To
make disordered ice, you can cool water rapidly鈥攁t a million degrees a
second. Or squeeze ordinary ice to 10 000 atmospheres at a temperature of
鈭�196 掳C. It turns out that the high-pressure route produces a
high-density amorphous ice, while rapid cooling makes a low-density amorphous
ice (see Diagram).
So there are two kinds of glassy ice, and one has a higher density than the
other. Perhaps, then, these glassy ices might be 鈥渁rrested鈥� analogues of the HDL
and LDL states of supercooled water?
One way to find out would be to melt the ices to get their molecules moving
again, and to study the resulting liquids. But this approach runs against
another barrier to the no-man鈥檚-land: the ices melt at around 鈭�120 掳C,
which is still way below water鈥檚 freezing point. So the instant they melt, they
solidify into ordinary ice, wiping out any chance to look at the liquid. In the
past year, however, Stanley and Osamu Mishima of the National Institute for
Research in Inorganic Materials in Japan have finally found a way to catch a
glimpse into the no-man鈥檚-land.
Dividing line
Their idea was to map out the topography of the 鈥淕ibbs potential surface鈥� for
water in the supercooled region. This is a sloping mathematical surface that
reflects the relationship between water鈥檚 temperature, pressure and density.
Sharp kinks in the surface signify phase transitions between different states.
To do the mapping, the researchers started with droplets of 鈥淚ce IV鈥�, a
high-pressure crystalline form of ice, which they then melted by altering the
pressure. The changing temperature of the sample traced out a path across the
surface. 鈥淎ll we did鈥�, says Stanley, 鈥渋s squeeze on the sample and measure the
迟别尘辫别谤补迟耻谤别.鈥�
In this way, Mishima and Stanley strode up and down the Gibbs potential
surface measuring its shape. And at the dividing line where they thought the two
liquid forms of supercooled water would coexist, they found a telltale
kink鈥攁 sure sign of a phase transition. They couldn鈥檛 observe either
liquid directly, but they felt their combined imprint on the Gibbs surface.
But there鈥檚 no need to worry next time you take a bath that the waters might
suddenly separate. Nor should you expect a better selection of liquids at the
drinking fountain. 鈥淲e鈥檒l never have something to pass around the room,鈥� notes
Stanley. Water might be two-faced鈥攂ut we may never see either of them in
the flesh.