The formation of crystals is one of the most commonplace processes you can probably think of. Every time you freeze water into ice cubes, for instance, you're creating crystalline structures. There's even a fun experiment you can do to grow salt crystals – with nothing more than table salt and water.

But on the atomic level, we have a poor understanding of how crystals form, particularly nucleation – the very first step in the crystallisation process. That's partially because it's a dynamic process that happens on such small scales, and partially because it's somewhat random, both of which make it difficult to study.

That's what makes the work of a team of researchers led by chemist Takayuki Nakamuro of the University of Tokyo in Japan so exciting. Using a special technique in development since 2005, they have filmed the crystallisation of salt on the atomic scale for the first time.

Since crystallisation is used for a wide number of applications - from medicine to industrial manufacturing – this is a step towards better controlling how we create materials, the researchers said.

The technique is called single-molecule atomic-resolution real-time electron microscopy, or SMART-EM, used to study molecules and molecular aggregates. By combining it with a newly developed sample preparation method, the team captured the very formation of salt crystals.

salt crystals(The University of Tokyo)

"One of our master's students, Masaya Sakakibara, used SMART-EM to study the behaviour of sodium chloride (NaCl) - salt," Nakamuro said.

"To hold samples in place, we use atom-thick carbon nanohorns, one of our previous inventions. With the stunning videos Sakakibara captured, we immediately noticed the opportunity to study the structural and statistical aspects of crystal nucleation in unprecedented detail."

At a rate of 25 frames per second, the team recorded as water evaporated from a sodium chloride solution. From the liquid chaos, induced by the shape of a vibrating carbon nanohorn suppressing molecular diffusion, order emerged as tens of salt molecules emerged and arranged themselves into cube-shaped crystals.

These pre-crystallisation aggregates had never been observed or characterised before, the researchers said.

Nine times the researchers observed the process, and nine times the molecules arranged themselves into a cluster fluctuating between featureless and semi-ordered states before suddenly forming into a crystal: four atoms wide by six atoms long. These states, the team noted, are extremely different from the actual crystals.

They also noticed a statistical pattern in the frequency at which crystals formed, grew and shrank. They found that during each of the nine nucleations, the timing of the nucleation process roughly followed a normal distribution, with an average time of 5.07 seconds; this had been theorised, but this is the first time it has been experimentally verified.

Overall, their results showed that the size of the molecular assembly and its structural dynamics both play a role in the nucleation process. Understanding this, it is possible to precisely control the nucleation process by controlling the space in which it occurs. They could even control the crystal size and shape.

The next step in the research will be to try and study more complex crystallisation, with broader practical applications.

"Salt is just our first model substance to probe the fundamentals of nucleation events," said chemist Eiichi Nakamura of the University of Tokyo.

"Salt only crystallises one way. But other molecules, such as carbon, can crystallise in multiple ways, leading to graphite or diamond. This is called polymorphism, and no one has seen the early stages of the nucleation that leads to it. I hope our study provides the first step in understanding the mechanism of polymorphism."

The research has been published in the Journal of the American Chemical Society.