MAKING new materials has traditionally been a matter of good cookery. Choose your ingredients, mix them carefully, pop them in the oven at the right temperature and the result – if you are lucky – is a new plastic, ceramic, semiconductor, superconductor, or whatever. But for a new generation of materials, this mix and bake approach is being replaced by architectural design. żěè¶ĚĘÓƵs are developing a growing repertoire of molecular modules that can be made to assemble themselves into highly regular materials. They could eventually find a range of uses from antibiotics to industrial catalysts. The key is the glue that holds the modules together – subtly powerful connections called hydrogen bonds.
There is nothing new about the power of hydrogen bonds. The water molecules in ice are held together by them – and a sheet just a few centimetres thick will support a fully grown skater. In spider silk, hydrogen bonds hold protein molecules in sheet-like assemblies that are aligned in strands stronger than steel. In tendons, they secure molecules of the protein collagen into stiff coils. But behind this impressive collective strength lies individual weakness. Hydrogen bonds are feeble compared with the familiar covalent bonds that bind atoms to each other in molecules, or the ionic bonds that hold salts and minerals together.
Surprising as it seems, this weakness can be a virtue. In some situations, covalent and ionic bonds are just too strong. Take the folding of protein molecules into their biologically active shape, for example. This complex process involves step after step of trial and error: the protein chain must be gently shuffled into place, then unravelled again whenever the folding goes wrong. If each twist and turn was secured by a covalent bond, errors in folding would be frozen in and the protein would end up defective. But molecular building blocks that are held in place by an array of weak forces can be prised apart and reworked so that even complex structures can be virtually free of defects.
Advertisement
The second advantage of hydrogen bonds is that, unlike ionic bonds, they point in a definite direction in space. So using them to assemble molecular structures is like building toy models using bricks that will join to each other only in specific ways, rather than like stacking oranges. And instead of laboriously building up the structures brick by brick, it sh6uld be possible to fabricate the molecular building blocks with suitably positioned “sticky ends”, ready to form hydrogen bonds that will guide the blocks to assemble themselves into a particular desired structure.
Reza Ghadiri and colleagues from the Scripps Research Institute in La Jolla, California, have done just that. In late 1993, Ghadiri and his team announced that they had made molecular tubes less than a nanometre in diameter which had assembled themselves spontaneously from cyclic peptides – synthetic, ring-shaped molecules that, like proteins, are made up of a string of amino acids. The peptides were designed to stack on top of each other like hoops, each one bound by hydrogen bonds to its neighbours above and below.
One problem Ghadiri’s team faced was that most peptide molecules have a tendency to crumple up. They solved this by using both left- and right-handed amino acids when they built the ring. Every amino acid (with the exception of the simplest, glycine) comes in two varieties, which are identical apart from being mirror images of each other. Alternating the two forms gives the ring a puckered shape that is much less likely to collapse.
The bonds that connect amino acid units together are called peptide bonds. Each peptide bond contains an oxygen atom, which can form one half of a hydrogen bond, and a hydrogen atom attached to nitrogen, which can act as the other half. In Ghadiri’s puckered peptide rings, these two atoms point upwards and downwards, more or less perpendicular to the plane of the ring – just right for them to forge hydrogen bonds linking rings in a vertical column.
The final design element in Ghadiri’s cyclic peptides was a way of triggering them to self-assemble on command into tubelike structures. For some of the left-handed amino acids, the team used glutamic acid, which has a side chain containing a carboxylic acid group. In alkaline solutions this loses a positively charged hydrogen ion to become a negatively charged carboxylate group. Ghadiri reasoned that the negative charges on the rings would make them repel one another and prevent self-assembly. In more acid solutions, the carboxylates would get their hydrogen ion back, neutralising their charge and triggering assembly of the tubes.
This turned out to be precisely what happened. When the Scripps researchers acidified a solution of the cyclic peptides, long needle-like crystals appeared. Viewed under an electron microscope, these crystals appeared to have striations along their long axis – a tantalising clue that the crystals were built up from an array of tubular molecules. After many painstaking attempts, the researchers finally produced crystals large enough for analysis by X-ray diffraction, which showed up the precise positions of the atoms. The rings were indeed stacked into tubular channels (see Diagram).
Porous materials that contain channels about the same size as individual molecules are much in demand for a range of applications. The microporous minerals called zeolites have an illustrious history as industrial catalysts for reactions that rearrange and break down hydrocarbons, such as those involved in cracking crude oil. Zeolites are so useful because their networks of molecular-scale pores, channels and cavities make them highly selective catalysts. Molecules too large to fit into the pores are excluded from the reactions that take place on the pore walls. The same property also makes zeolites useful as molecular filters for chemical separation. Industrial chemists would dearly love to be able to produce artificial zeolites with a specified pore size and channel geometry. But making such materials is a hit-and-miss affair, in which little control of the pore shape is possible. Ghadiri’s peptide tubes suggest that a building-block approach may provide this control. If the cyclic peptides could be designed with catalytic molecular groups attached, the self-assembled tubes might then act as size-selective catalysts.
Ghadiri’s tubes might also be used as antibiotics. Some natural molecules whose normal function is to help transport metal ions across internal cell membranes will kill bacteria by punching holes their cell walls. For example if the peptide gramicidin A is incorporated into a membrane, it forms a coil; two such coils end to end span the membrane, creating a cylindrical channel through which ions can pass.
In May last year, the Scripps group found that its synthetic cyclic peptides acted as efficient ion transporters. The researchers incorporated the peptides into artificial membranes, which were separating two solutions in which the concentration of hydrogen ions, or pH, was different. They found that the peptides allowed the pH to equalise by encouraging hydrogen ion transport through the membrane. They believe the molecules assemble themselves into the same tubular stacks that they saw in the crystals, providing channels through the membrane like those formed by gramicidin A.
In the artificial membrane systems, the synthetic peptide tubes transport ions three times as fast as gramicidin A. This raises hopes that such self-assembling systems have great potential as antibiotics. Moreover, as they are held together only by hydrogen bonds, it might be possible to trigger the channels to assemble or dismantle themselves on demand, and so to switch their antibiotic activity on and off. Ghadiri and his team have also been able to make wider channels, which allow glucose molecules to pass through membranes. This might ultimately prove useful for treating diabetes, which is caused by a deficiency in glucose transport across cell membranes.
Other groups are also getting in on the act. Last October Jeffrey Moore and colleagues at the University of Illinois at Urbana-Champaign announced that they too had made a porous material from ring-shaped units that stack themselves by hydrogen bonding. Moore’s molecular rings are rather different from Ghadiri’s. Each ring is made up of six rigid rod-like units and six corner pieces to form a regular hexagon. For the rods, they used acetylenic groups – two carbon atoms joined by a triple bond into a linear unit. The corner pieces were benzene rings, which provide connections at exactly the 120° angle needed to make hexagons. The resulting “macrocyclic” hexagons are very stiff, like washers.
Molecular sieve
To encourage the macrocycles to stick together, Moore gave them hydrogen bonding units at every corner, in the form of hydroxy groups attached to the benzene rings. Unlike Ghadiri’s cyclic peptides, however, these hydrogen bonding groups do not direct the stacking of one ring on top of another. Instead, the “sticky” groups point outwards, in the plane of the ring, so the rings should assemble themselves into a hexagonal array, forming large, flat sheets (see Diagram). Moore’s team discovered that the benzenes in neighbouring sheets interact in a way that encourages the large rings to stack more or less on top of one another, creating parallel channels about 9 angstroms in diameter. Moore hopes that such an assembly might act as an organic molecular sieve.
A slightly different approach has been developed at the University of Montreal in Canada by Jim Wuest’s group. Rather than starting with preformed pore-forming building blocks like Ghadiri’s peptide rings and Moore’s rigid hexagons, Wuest uses blocks that create the channels only as they assemble themselves into a network. The blocks form the corners of the network, within which there are a series of interconnecting voids. Wuest calls his building blocks “tectons” (from the Greek word meaning “build”), and to make them self-assemble he gives them sticky extremities that can form hydrogen bonds. This provides control over the direction in which the bonds form, so the geometry of the tectons determines that of the network into which they assemble.
Earlier this year Wuest’s researchers showed that a tetrahedral tecton with hydrogen bonding groups at each of its four vertices will self-assemble into an arrangement of interconnected tetrahedra known as a diamondoid network. Diamond is made up of a network of carbon atoms, in which each atom is linked to four others positioned at the corners of a regular tetrahedron.
Wuest’s crystalline material is similar, except that the units are not individual atoms but molecular tectons (see Diagram). The diamondoid network has large amounts of empty space between the connecting arms, and it is these voids which interconnect to form channels.
But Wuest’s scaled-up diamond is a little more complicated than this. In general, crystal structures are most stable when the units that make them up are packed densely, because this maximises the attractive interactions between them. So as Wuest’s materials form, they cut down the amount of empty space by forming not just one diamondoid network but two, delicately interwoven so that the units that form one network fill part of the void space of the other. Yet even this seemingly crowded arrangement leaves a system of square channels 6 angstroms across running through the network.
As these channels form, they trap molecules of solvent. If the crystal is immersed in another liquid the molecules of the liquid can take the solvent’s place. But if these entrapped “guest” molecules are expelled without being replaced by others, the network collapses, revealing that the diamondoid structures are far less stable than zeolites. Wuest hopes to find ways to make them more stable by forging stronger cross-links between the tectons once they have self-assembled by hydrogen bonding.
It is not only three-dimensional structures that can be made using this approach. Earlier this year, Toyoki Kunitake and colleagues from Kyushu University in Japan described how polymer-like threads form from a solution containing two different kinds of molecule, one based on the compound melamine and the other a type of compound called a diimide. These two molecules have hydrogen bonding groups arranged around their edges in just the right places to allow them to slot together as alternating melamine-diimide chains.
Acid partner
This research builds on extensive exploration of melamine’s hydrogen bonding abilities by George Whitesides and his group at Harvard University. Whitesides has matched melamine with a partner based on barbituric acid, which contains complementary hydrogen bonding groups. Shutting off some of the hydrogen bonding capacity can give different-shaped assemblies. For example he has managed to make a zigzag polymer tape. Polymers in which the molecular units are held together by weak bonds, rather than by a covalently bound backbone, could turn out to have some really valuable properties. Concentrated solutions of chain-like molecules can be very viscous, because the chains get tangled. But if the components of the chains are only weakly linked, stirring them might be enough to sever the chains, causing the polymer to flow freely. When the stirring stops, the sticky edges should ensure that the polymer can reassemble into a thick tangle of strands. Nondrip paint works on a similar principle, but at present the polymers used have covalent backbones, and the strands are held together by other weak attractive forces until the paint is stirred.
All this suggests that hydrogen bonding could prove to be an invaluable tool for researchers in their quest for designer materials. So far, the focus has been on learning the ropes – discovering how to control the way molecules stick together. The next step will be to incorporate active components -catalytic groups, perhaps, or components that emit or respond to light, like the motors and light bulbs in toy construction kits. Then we will really see these new molecular materials in action.
Nature’s glue
HYDROGEN bonds are the result of electrostatic attraction between electron-rich and electron-poor regions of molecules. When a hydrogen atom is attached by a covalent bond to oxygen or nitrogen it acquires a slight excess positive charge. This is because such atoms pull electrons away from the hydrogen. The slightly positive hydrogen may then be attracted to regions of high electron density – and hence negative charge – such as the “lone pairs” of nonbonding electrons belonging to other oxygen and nitrogen atoms. The force of electrostatic attraction between a lone pair and a positively polarised hydrogen atom is what forges the hydrogen bond. It has only about one-tenth the strength of a typical covalent bond.
This feebleness is put to good use in biology. For example, hydrogen bonds bind together the two strands of the double helix of DNA. They are strong enough to hold the strands together, so that it provides a reliable molecular databank. But they can be gently unzipped for transcription, the first step in converting a gene to a protein, and when the DNA is replicated as the cell divides. The separated strands can be reunited smoothly even after they have been pulled apart completely. Hydrogen bonds also help to keep the amino acid chains of protein molecules folded into their characteristic, biologically active forms.