explainers

The Enduring Mystery of How Water Freezes

Making ice requires more than subzero temperatures. The unpredictable process takes microscopic scaffolding, random jiggling and often a little bit of bacteria.

New simulations indicate that ice crystallization happens fastest — this slow-motion movie covers mere nanoseconds — when water is tuned to a critical point called the liquid-liquid transition.

Introduction

We learn in grade school that water freezes at zero degrees Celsius, but that’s seldom true. In clouds, scientists have found supercooled water droplets as chilly as minus 40 C, and in a lab in 2014, they cooled water to a staggering minus 46 C before it froze. You can supercool water at home: Throw a bottle of distilled water in your freezer, and it’s unlikely to crystallize until you shake it.

Freezing usually doesn’t happen right at zero degrees for much the same reason that backyard wood piles don’t spontaneously combust. To get started, fire needs a spark. And ice needs a nucleus — a seed of ice around which more and more water molecules arrange themselves into a crystal structure.

The formation of these seeds is called ice nucleation. Nucleation is so slow for pure water at zero degrees that it might as well not happen at all. But in nature, impurities provide surfaces for nucleation, and these impurities can drastically change how quickly and at what temperature ice forms.

For a process that’s anything but exotic, ice nucleation remains surprisingly mysterious. Chemists can’t reliably predict the effect of a given impurity or surface, let alone design one to hinder or promote ice formation. But they’re chipping away at the problem. They’re building computer models that can accurately simulate water’s behavior, and they’re looking to nature for clues — proteins made by bacteria and fungi are the best ice makers scientists know of.

Understanding how ice forms is more than an academic exercise. Motes of material create ice seeds in clouds, which lead to most of the precipitation that falls to Earth as snow and rain. Several dry Western states use ice-nucleating materials to promote precipitation, and U.S. government agencies including the National Oceanic and Atmospheric Administration and the Air Force have experimented with ice nucleation for drought relief or as a war tactic. (Perhaps snowstorms could waylay the enemy.) And in some countries, hail-fighting planes dust clouds with silver iodide, a substance that helps small droplets to freeze, hindering the growth of large hailstones.

Konrad Meister, a biophysical chemist at Boise State University in Idaho, uses a video wall to explore the structure of ice-nucleating proteins.

Courtesy of Konrad Meister

But there’s still much to learn. “Everyone agrees that ice forms,” said Valeria Molinero, a physical chemist at the University of Utah who builds computer simulations of water. “After that, there are questions.”

Freezing Water

What’s special about zero degrees is that, at or below this temperature, it makes energetic sense for water to turn from liquid into ice. Below that threshold, ice’s crystal structure has a lower energy than sloshing water molecules. The process of freezing water actually releases heat, which is why you can use an infrared camera to see ice heat up as it solidifies.

Ice nucleation begins when, by chance, random jiggling arranges a small patch of triangular H2O molecules into a hexagonal ice structure. This ice embryo might grow into a nucleus and kick off freezing. Or it might just dissolve away. That’s because there’s an energy barrier that keeps embryos from growing. It’s energetically costly to form an interface between ice and water; molecules arranged in an ice structure butt up against molecules of the surrounding liquid, and the resulting imbalanced forces make the interface unstable. Until a speck of frost reaches a certain size, the cost of its interface overwhelms the payoff of energy released by the ice formed inside.

This barrier to nucleation is like being at the top of a sea cliff on a hot day, said Konrad Meister, a biophysical chemist at Boise State University who studies biological antifreezes and ice-nucleating substances. You’re hot; you’d rather be in the water. But without a gust of wind to push you off or a friend to encourage you to jump, your fear keeps you paralyzed, trapped in the less-ideal state at the top of the cliff.

A snow gun sprays water into the air amid a snowy scene.

Snow guns used at ski resorts spray water into the air mixed with an ice-nucleating agent, often proteins from the bacterium Pseudomonas syringae.

zedspider/Shutterstock

The colder that water gets, the smaller this energy barrier gets. This makes it easier for random molecular motions to push a tiny embryonic ice structure over the critical size threshold. Ice forms and grows, and the lower-energy crystal structure stays stable.

Boosting Nucleation

Surfaces and impurities can dramatically lower the energy barrier for nucleation — and therefore raise the temperature at which ice forms. “Since the late 1970s, we’ve known that there are lots of aspects of a surface that are important,” said Miriam Freedman, an atmospheric chemist at Pennsylvania State University.

Like a microscopic construction scaffold, surfaces with the right structure make it easier for water molecules to arrange themselves into a crystal. Researchers have identified a few things that can make a surface better or worse at nucleating ice. A surface’s crystallinity, or structural orderliness, matters. And substances with chemical structures that mimic ice tend to be good at ice nucleation. Pores of a certain size confine water molecules in a way that helps ice form, too.

Meister and Molinero have been working together to unravel the secrets of nature’s best snow makers — bacteria and fungi whose proteins interact with water in ways that promote ice nucleation. Many of these organisms are plant pathogens, and it’s possible that their ice-nucleating proteins evolved to cause frost damage.

Valeria Molinero, a physical chemist at the University of Utah, builds computer simulations of water to study ice nucleation.

University of Utah

The best known ice nucleator is a bacterium called Pseudomonas syringae, which has a protein that can force water to freeze at around minus 2 degrees Celsius. “It’s so good that all the artificial snowmaking, at least in Utah, and some [other] places in the U.S., uses this bacteria to make snow,” Meister said.

Bigger proteins tend to be better for making ice, possibly because they act as a more effective template: Imagine trying to build a skyscraper with a scaffold just a few stories tall.

But with all they know, scientists still encounter surprises. Meister, Molinero and their co-authors recently discovered an exception to the bigger-is-better rule: fungal proteins that are great at ice nucleation despite being tiny. They get around the problem by clumping together into large, ice-nucleating aggregates.

Predicting Ice

Molinero develops theories and computational models that capture how ice nucleates, including its interaction with surfaces.  In 2009, she and her colleague Emily Moore published a simplified model of water that treats each H2O molecule as a single, tetrahedron-shaped atom; surprisingly, computer simulations of this monatomic-water model accurately reproduce water’s large-scale properties, like its density. Then, in 2011, Molinero and Moore used the monatomic-water model to pinpoint a specific structural change in supercooled water that sets the lower limit of water’s freezing point. The model predicts that water must freeze at minus 48.15 C.

More recently, in computer simulations published in May in the Proceedings of the National Academy of Sciences, Molinero and her colleagues showed that ice crystallization happens fastest when water’s temperature and pressure are tuned to a point of transition between denser and less dense liquid phases. And in March, they presented a new model at the American Chemical Society conference that can predict the temperature at which ice will nucleate on a given surface. The model is informed by experimental data and considers a battery of factors, from the surface’s chemistry to the shapes of its defects.

Depending on their size and geometry, bumps and divots on a surface can squeeze water molecules into configurations that make it easier or harder for ice to form. As part of their model, Molinero’s team developed and tested a new formula for how the bump or divot’s angle affects ice nucleation. Using the formula, Molinero thinks it should be possible to design better ice-nucleating materials just by introducing defects of the right size and shape. “You can take a surface that is not so good and make it quite outstanding,” she said.

According to Molinero, the models atmospheric scientists use to predict cloud behavior don’t yet account for the nuances of ice nucleation. And it’s still unclear which particles are actually the most important for seeding clouds in nature. Mineral particles like Saharan dust are abundant in the atmosphere and can nucleate ice. But they’re not alone up there.

“Up in the clouds, you find some of these bacteria, some of these fungi, that are very good at ice making,” Meister said. “That completely raises the question: What makes it rain?”

Comment on this article