In a chemical reaction, participating molecules swap components and rearrange their shapes, sometimes taking on forms that last for only a fraction of a second before the final reaction products are formed. Because of the transient, unstable nature of these reaction intermediates, it has been difficult to catch a glimpse of their three-dimensional structures.

But Makoto Fujita and his colleagues at the University of Tokyo may have found a solution. Using a textbook chemical reaction as an example, on page 633 the chemists show how they slowed down the reaction process by trapping the reagents in 'crystalline molecular flasks'. This, in turn, allowed them to capture snapshots of an elusive intermediate by X-ray crystallography.

Fujita had the idea of assembling crystalline molecular flasks in which to carry out chemical reactions several years ago. His team's first attempt consisted of chemical 'cages' crystallized together with a substrate to render that reagent more stable and resilient. “The idea worked quite well, and we carried out some chemical transformations in the cavity of the cage,” says Fujita.

But the approach turned out to be impractical: “Having to crystallize the cage for each reaction was not always easy and normally time consuming,” says Fujita. To remedy this, he assembled a network of connected cages, or pores, thereby forming a crystalline structure that was stable over a range of temperatures.

“The porous network has a big advantage over the cage because it does not require crystallization each time,” says Fujita, who published the first attempt at assembling the network in 1994. By this method, the structure exists as a crystal to which various substrates can simply be added, depending on the reaction being studied.

To test whether the porous network could be used to perceive chemical reactions by X-ray crystallography, Fujita and his colleagues chose a well-known organic reaction between an amine and an aldehyde. The reaction, known as Schiff-base formation, produces a hemiaminal intermediate that is normally very short-lived.

Fujita and his colleagues assembled their porous network embedded with amine, and then immersed it in acetaldehyde. Once both molecules met in a single pore, the reaction took place — but at a slower pace than normal. “The reaction times were retarded within the pore because of steric [space] restrictions. Reactions could also be paused at any time by cooling the crystal to liquid-nitrogen temperature, explains Fujita, which enabled the team to obtain structural data for the paused reaction steps.

The biggest challenge, he says, was to optimize the reaction conditions, such as crystal quality, solvent, temperature and reaction time. In particular, they had to tweak conditions so that the concentration of the hemiaminal intermediate was more than 35% within the porous network. Once achieved, “We finally succeeded in observing hemiaminal by X-ray crystallography”, he says.

Although the structure of the hemiaminal intermediate was not itself that surprising, the study demonstrates a useful approach for analysing chemical-reaction intermediates. Fujita is now forging ahead to examine more unusual structures, such as the intermediates formed during organometallic transformations or enzymatic reactions with transition-metal ions, some of which, he says, have never been observed.