ONE of the exhibits at London’s Science Museum is a load of old balls. Stroll through the ground floor galleries, go up a staircase to the right, and there you will find it: in a long glass case, together with a manually operated fire engine and a digger built for open-cast mining.
The display tells the story of the making of the modern world, and this particular exhibit represents one of its most revolutionary developments. It is a model built by chemist Dorothy Crowfoot Hodgkin of the penicillin molecule, the first antibiotic to be discovered. The multicoloured balls joined with little sticks represent the 3D arrangement of atoms inside the molecule – a secret that, once uncovered, enabled us to mass produce the drug and save millions of lives.
Advertisement
It’s easy to see how the model might be a bit outdated in a world where supercomputers can work out the configuration of atoms with fine precision. But it’s not just the model itself that belongs in a museum. Our entire conception of molecules is increasingly coming to seem like a period piece. Move beyond this crude view of little balls and sticks, and it could shift chemistry up a gear, opening a path to discovering new types of drugs and smarter, faster ways to make the materials that the world depends on.
To understand the importance of manipulating molecules, you need only look at the plastic casing of your smartphone, or the flecks of toothpaste on your polyester shirt. There is hardly an area of modern life that doesn’t rely on the elegant craft of stitching simple molecules into newer and more interesting ones. But the story of penicillin really illustrates its power.
During the second world war, chemists made the wonder drug by culturing moulds in pans. Labs would have these pans stacked to the ceiling, but the process still yielded only tiny amounts of penicillin. The compound was in such short supply that unmetabolised remnants of the drug were being recycled from soldiers’ urine. To work out a better way of making it, we first needed to know what its molecules looked like – how the atoms were connected.
That was where Hodgkin and her model came in. She worked out the molecular structure by shining X-rays through crystals of penicillin and interpreting the patterns that emerged. In 1945, she made the model that now sits in the . Chemists could then begin to think about building up the molecule it represented, a feat that John Sheehan at the Massachusetts Institute of Technology eventually pulled off in 1957.
Even Hodgkin and Sheehan knew that molecules aren’t actually miniature balls and sticks. We have long known that bonds stretch and atoms vibrate. Atoms can even pull off quantum tricks like disappearing and appearing again – difficult to replicate with beads and pipe cleaners.
But the simplified model remains surprisingly handy. After all, chemical reactions are all about breaking and making covalent bonds, the sticks between two balls. These bonds are nothing more than two electrons shared between two atoms. Sometimes the bond can split so that each atom gets one electron, producing highly reactive fragments called radicals. But for the most part, just before the molecules fall apart the bonds stretch out into a severely strained posture before finally breaking and forming new bonds to create a product.
That contorted position, halfway between the starting molecule and the product, is called a transition state, and molecules don’t like it one bit. That’s because it takes a lot of energy to get there. Think of it as a bit like the top of a playground slide, a halfway house you need to pass through in order to reach the bottom. Transition state theory says that if you know the ball-and-stick structures at each of those three stages, you can calculate the energy losses and gains involved in climbing and whizzing down the slide. This tells you how quickly the reaction will go – a crucial piece of information if you are planning on making a particular chemical in the lab. All you need to make an accurate prediction is the location of each atom. Chemists couldn’t ask for a better theory: it is simple and intuitive, and it always works.
That is, until it doesn’t. In 2004, rumours crept along the academic grapevine that certain reactions were passing through odd transition states, suggesting a gap in the theory. One was the decomposition of formaldehyde, a carbon bonded to an oxygen and two hydrogens. Best known for its use in embalming fluid, formaldehyde is also common in the manufacture of plastics and explosives. Could an exception to the theory have been hiding in plain sight all along?
The idea intrigued chemist Arthur Suits, then at Stony Brook University, New York. He knew he had to move fast to be the first to get to the bottom of it, but he was getting ready to take a position at another university halfway across the country. With his laboratory about to be thrown into chaos, he and his team set about isolating individual formaldehyde molecules in the lab and watching how they decomposed into carbon monoxide (CO) and hydrogen. His equipment gave him the tools to trap the CO fragment and measure its energy. That in turn allowed him to predict the energy of the hydrogen molecule, because the two figures must add up to the energy of the formaldehyde molecule.
“The theory is simple and intuitive, and it works – except where it doesn’t”
“It was down to the wire,” says Suits. “We stayed all night in the lab.” The experiments worked, revealing that the CO fragments were rotating incredibly slowly. That meant the hydrogen molecule must have had a huge amount of energy, almost enough for the hydrogen atoms to have broken free on their own. “That didn’t fit with the picture,” says Suits.
Suits called Joel Bowman, a chemist at Emory University in Atlanta, Georgia, who had been running a computer simulation of the same reaction. Bowman could see the other half of the picture, the detailed behaviour of the hydrogen, and was getting results that matched Suits’s calculations.
In Bowman’s simulation, the hydrogen atoms were misbehaving spectacularly. Instead of running through a transition state, one hydrogen atom appeared to stretch away from the rest of the formaldehyde molecule before settling into an , eventually joining up with the other atom on the other side and zipping off together.
The reaction was unlike anything anyone had seen before. Instead of snapping in two or contorting awkwardly, formaldehyde seemed to choose a third way of decomposing: allowing one of its constituent atoms to go walkabout (see “The third way”).
“We spent a long time thinking about the right word for this,” says Bowman. “We eventually settled on the word ‘roaming’.”
Roaming soon showed up in five or so other simple compounds, demonstrating the limitations of the static ball-and-stick way of thinking. The more complicated jigglings and wagglings that chemists had ignored for decades, assuming they had no major impact on reactions, were now a hot topic. If the motion of atoms were not taken into account, chemistry would be incomplete. “We have to rethink our idea of what a reaction is and how it occurs,” says Bowman.
Chemistry in motion
Fourteen years later, not everyone sees roaming as a sea change. “People have been talking about roaming for a very long time,” says David Glowacki, a chemist at the University of Bristol, UK. He points out that apart from one or two controversial papers that stretch the definition of roaming, chemists have only ever observed the mechanism in near-vacuum conditions. “People argue about if these effects carry over into things we actually care about – like liquids – because that’s where most of the chemistry in the universe happens.”
In the past few years, however, we have started finding more and more examples of subtle molecular jigglings besides roaming that throw off the predictions of transition state theory – and these apply to liquid phase reactions that chemists use all the time for building molecules.
“This area is exploding,” says Dan Singleton at Texas A&M University. He has uncovered several such mechanisms, which he calls dynamic effects.
Years ago, chemists believed that molecules entering a transition state had only one way out, producing the same set of end products every time. One of the mechanisms Singleton has identified, however, throws this logic out the window. Think of a molecule climbing a ladder towards a transition state, but then having several different slides descending towards the ground to choose from. As conditions change, different slides become preferred, meaning that molecules in the same transition state can wind up taking different routes to the bottom. As a result, you end up with a ratio of products in your flask that is different to what transition theory predicts. “You run into these things in ordinary textbook reactions that every chemist learns as an undergraduate,” says Singleton, although the discrepancies have long been ignored.
The third way
Chemists thought that molecules could come apart in two ways, and that a static ball-and-stick model could depict them both. The realisation that atoms can move around when molecules like formaldehyde break up– a dynamic process called roaming– has forced a rethink. (Click here to see a static image of the explanation below)
Roaming
One of the hydrogen atoms goes into orbit around the formaldehyde molecule, before finding the other one and breaking off altogether
Conventional transition
The molecule contorts into a transition state, allowing the hydrogen atoms to form their own molecule
Radical dissociation
The bond holding one hydrogen atom snaps, leaving both fragments with an extra electron
The problems transcend the lab bench. Dean Tantillo at the University of California, Davis, has calculated that dynamic effects are involved when plants make a group of volatile chemicals called terpenes; some of these are responsible for that pine-fresh smell of forests. In this case, the multiple slides seem to help plants produce a huge variety of complicated molecules from a single reactant.
What does this all add up to? “The ball-and-stick picture is nice, but antiquated,” says Stephanie Hare, a graduate student in Tantillo’s lab. “Molecules are constantly wiggling, and thinking of bonds as hard, stiff connections between atoms makes it difficult to understand the dynamics of chemical reactions intuitively.”
If we could understand how dynamic effects work, we might be able to harness them to design better reactions, and perhaps find new ones that enable us to speed up synthesis. It might even lead to a new way of finding drugs.
The trouble is, with transition state theory out the window, we need some alternative way of understanding reactions. Stephen Wiggins, a mathematician at the University of Bristol, believes he may have just the thing. Along with Glowacki, he and others have scooped more than £4 million of funding from a UK research council to investigate ways of representing dynamic chemistry.
“We have to rethink our idea of what a reaction is and how it occurs”
The first step, says Wiggins, is to build descriptions of molecules that are not static, as in transition state theory, but take the atoms’ movement into account. This means describing each atom with not just its coordinates in three-dimensional space but with its speed and direction of movement too.
The trouble is, this approach doubles the amount of data needed to describe the molecule at any given moment, making simulations computationally demanding. Worse, if you change the initial conditions just slightly, the complexity of the model means you can get very different results. In short, no one quite knows what the solutions spat out by the simulations really mean.
Wiggins is just beginning to tackle this problem, using machine learning algorithms to spot patterns in the flood of data that we can use to predict how a reaction will proceed. The possibilities are immense. Understand how molecular vibrations determine the way reactions happen, says Wiggins, and we could turn that to our advantage. We might be able to tune reactions along specific pathways, forcing them to deliver the exact chemicals we want. Alternatively, we could find entirely new reaction mechanisms we had never imagined, with far-reaching implications for chemical synthesis. “It’s a very tough challenge,” says Singleton. “I’m not positive they’ll succeed – but they’re doing the right thing to try.”
Wiggins’s model – composed of lines of computer code and sophisticated mathematics – may not have quite the aesthetic appeal of the Science Museum’s physical models. But if his ambitions pay off, their redesign of chemistry might earn them the right to a glass case of their own.
Virtual chemistry
Getting to grips with chemical reactions would be so much easier if we could see the molecules at work.
That’s why David Glowacki at the University of Bristol, UK, has developed a virtual reality tool that allows chemists to picture the structures they study in the lab. In preliminary tests, Glowacki has shown that people are much faster at moving digital models of molecules around using VR gear than they are doing the same task on a screen.
Glowacki hopes to harness that spatial intuition to let humans tackle problems too hard for supercomputers to solve. He plans to set up a game that makes players manipulate molecules, all jiggling over time, in a race to find ones that fit together. The task has particular relevance to the pharmaceutical industry. “Drug design is a bit like four-dimensional Tetris”, says Glowacki.
In a way, Glowacki sees VR as an extension of the ball-and-stick models that chemists were using 70 years ago. Those were fully 3D, capable of instantly conveying a molecule’s geometry, but frustratingly static. In the 80s, chemists replaced them with digital representations of molecules that could depict motion but lacked valuable spatial information because they were displayed on 2D monitors. With Glowacki’s tool, the molecules can be intuitively viewed from all angles without sacrificing any of their inherent dynamism. “VR enables us to have the best of both worlds,” he says.
It could be just the tool we need to view a whole new world of chemistry.
This article appeared in print under the headline “Roaming free”
