water is weird

This article is from the March 20, 2001, issue of Red Herring magazine.

Water -- heavier when chilled, lighter when frozen, absorbing enormous heat with but a slight rise in temperature, the foundation of life, the most common of liquids, and the strangest. If only we knew how it worked.

"We still don't quantitatively understand the physics of liquid water," says Richard Saykally, a world-renowned chemist at the University of California at Berkeley. As a result, our best computer models don't simulate reactions in water, like the folding of a protein or the docking of a hormone and its target cell -- each a Holy Grail of biotech. Solving the mysteries of water would do for chemicals and pharmaceuticals what the wind tunnel did for aerospace: substitute fast, cheap calculations for slow, costly experiments. "We're talking about billions of dollars saved here," Mr. Saykally says.

Experimentalists like Mr. Saykally study how water actually behaves, at temperatures ranging from just above absolute zero to a few hundred degrees above its boiling point. Theorists, meanwhile, attempt to hone their computer simulations to match more closely the experimental observations. As increasingly powerful computers bring the two approaches into somewhat better agreement, scientists are learning that water is even weirder than they had thought.

LIQUID MYSTERY
Take virtually any liquid -- molten iron, for instance -- and freeze part of it into a solid; the solid will sink to the bottom. But ice floats, and the question is why. Indeed, water reaches peak density at 4 degrees Celsius, or around 40 degrees Fahrenheit.

"In school," says H. Eugene Stanley, a physicist at Boston University, "we learn that if water is in equilibrium with ice, the temperature must be zero Celsius. That's not true. The water at the bottom is not zero, but four Celsius, and the reason is that below four Celsius, the water starts becoming lighter, so the heavier four-Celsius water sinks to the bottom of the glass and just stays there."

Mr. Stanley describes this as the most remarkable of the "magical properties" of water, although there are plenty of others. There are, for instance, 5 different kinds, or phases, of liquid water, not to be confused with the 12 to 14 different phases of ice. Ice forms a crystal lattice, and each phase has its own structure. As a crystal, ice is as different from water as diamonds are from pencil lead. You can, for example, supercool water so that rather than freezing at 0 degrees Celsius, as it prefers to do, it will stay a liquid down to roughly -38 degrees Celsius. Water typically won't freeze without some impurities around which its molecules can begin to coalesce. For this reason, researchers who study supercooled water do so with the purest water they can get.

At -38 degrees Celsius, however, even the purest water spontaneously turns to ice. When that happens, "it does so with an audible bang, like a little bomb," says Austen Angell, a University of Arizona chemist who holds the world record for supercooling water. From -38 to -120 degrees Celsius, it's ice all the way, a temperature regime that Mr. Stanley calls "no-man's land," by which he means "no liquid." But below -120 degrees Celsius, it's possible to make what's known as ultraviscous water, a liquid as thick as molasses. Below -135 degrees Celsius comes glassy water, a solid having no crystal structure.

Most of water's strange properties stem from the peculiar bonds formed between neighboring H₂O molecules. The bonds are formed by the two hydrogen atoms, which stick out from the oxygen at an angle of exactly 106 degrees -- "Mickey Mouse ears," Mr. Stanley calls them, "with the two positive hydrogen atoms as the ears, and two little feet sticking out, which are the negatively charged pairs."

The bond angle doesn't allow water molecules to bind ears-to-feet. Instead, the left ear of one molecule goes to the foot of a second, and the right ear goes to the foot of a third. At any given moment, only a few water molecules are likely to be bound at both ears and both feet. Others will have only three bonds, and still others only two.

The result is hard to simulate because you can't treat every water molecule as identical. Nor can you portray them as spheres, with perfect symmetry that would cut back on the number of spatial relationships, considerably easing the calculating load. Moreover, the electromagnetic forces between Mickey's ears and feet have a relatively long range, so you have to take into account not merely neighboring molecules but those farther apart as well.

TESTING THE WATERS
As computing power has grown exponentially over the years and modeling techniques have improved, so have simulations, which can now do a reasonable job of modeling a few thousand water molecules at a time. The models explain the four-degree temperature anomaly and some other conundrums just as Mr. Stanley did in his suggestion 20 years ago -- at any one time, the water molecules are engaged in the largest possible number of "good" hydrogen bonds. In ice, for instance, the hydrogen bond network is fully engaged, with each Mickey Mouse molecule locked onto its neighbors by two ears and two feet and occupying its maximum volume.

In water, because one or more of the hydrogen bonds is always broken, the molecules can move a little closer than they can in ice, allowing them more ways to arrange themselves. Lower the temperature, and you get "a little bit of a solid phase inside the liquid phase, and, as you lower the temperature further, you get more and more of these little bits of ice forming," Mr. Stanley says, like "plums in the plum pudding."

But does this transition between phases happen in reality or just in the computer? The ultimate test of a model is whether it predicts a phenomenon that experimentalists have yet to discover. In the case of Mr. Stanley's water simulation, this happened in 1992, when he and two collaborating physicists, Peter Poole of the University of Western Ontario and Francesco Sciortino of the University of Roma La Sapienta, noticed a coalescing of the plums in the plum pudding at roughly -50 degrees Celsius. The water seemed to be separating into a less dense phase of highly bonded water and a denser phase of less well-bonded water -- a kind of liquid water never before seen.

The proposition was, and still is, controversial. Indirect evidence is mounting, but direct evidence is hard to come by. "Heat capacity, compressibility -- quite a lot of the properties of water measured in that region show this type of divergence," says Mr. Saykally. "That's the standard hallmark of a phase transition near a critical point in the neighborhood."

The only direct experimental evidence of the phenomenon comes from Osamu Mishima of Japan's National Institute for Research in Inorganic Materials. In 1994, Mr. Mishima demonstrated that glassy water has high- and low-density phases and a transition from the former to the latter that Mr. Stanley says "pops like popcorn" as the glassy water expands. More recently, Mr. Mishima and Mr. Stanley have plotted the melting temperature against the pressure of superpure water and discovered kinks in the resulting curves -- kinks that are consistent with transition to a new form of liquid water.

Meanwhile, Mr. Saykally wants to infer Niagara Falls from a drop of water by fully calculating the behavior first of two water molecules, then three, four, and onward. He hopes to end up with a water model that is demonstrably better than that of Mr. Stanley or, for that matter, anyone else. "It should be able to do everything," Mr. Saykally says, "to calculate any properties whatsoever of liquid water more accurately than they've ever been described before."