Graphene's growing family: The successor to silicon turns up some interesting new materials

4 mins read

Since its discovery, graphene – the atom thick carbon material – has been hailed as the heir apparent to silicon. It has much to commend it; an unusual chemical structure means electrons move freely along the plane of the graphene sheet, encountering practically no resistance. But that's also where graphene's problems start. Unlike silicon, it is simply too good a conductor to operate effectively as a switch without some circuit design and materials processing gymnastics.

A second problem is that graphene is not easy to work with. Conventional semiconductor manufacturing processes will not deposit neatly patterned atom thick lattices and graphene does not stick to things readily. Work on directed self assembly, using other organic molecules such as DNA, may provide a way to steer pieces of graphene into place on a wafer substrate but, even then, graphene pieces may need to be treated chemically to provide better adhesion. That may, in turn, disrupt the properties that engineers desire. Rather than warping graphene into something that behaves more like silicon, some believe its future is as a complement to traditional metal oxide semiconductor (MOS) field effect transistors. For example, the BISFET developed at the University of Texas uses quantum-mechanical interactions between layers of graphene. Unusually for quantum-mechanical devices, which normally need to be cryo-cooled, BISFETs should work at room temperature with switching energies of a fraction of an attojoule – four orders of magnitude better than any CMOS device. But, because the spacing between the layers is crucial, making a BISFET will be even harder than making a pure 2D device. The secret to graphene's performance lies in the way the carbon atoms bond with each other. Molecules such as graphene and benzene, as well as nanotubes and buckyballs, are rings or lattices made, notionally, of alternating single and double bonds. However, that differentiation does not exist in nature. If the lattice is flattened, the molecular orbitals the electrons occupy overlap with those from neighbouring atoms. The result is a unified orbital that is, potentially, as large as the molecule itself. In effect, a ring of electrons forms above and below the flattened lattice (see fig 1). The energy of this arrangement is so low that it takes very unusual structures to stop the lattice from flattening. As a result, flat sheets and ribbons of carbon are extremely common in nature and are key to the structure of graphite and numerous important biomolecules such as chlorophyll, where delocalised electrons allow charge to move swiftly from end of the molecule to the other. Although chemists routinely talk of delocalised electron rings, physicists tend to look at the unusual orbital arrangement of the electrons as 'Dirac cones'. These are zones where the electron has no effective mass. At first, carbon seemed unusual in allowing these Dirac cones to form. But other materials have been found that show the same behaviour when structured in the right way, and which can deploy other novel properties such as the ability to act as insulators or highly efficient conductors depending on the direction and spin of the electrons. One leading candidate, not least because its chemical symbol name suggests it could follow silicon MOS, is molybdenum disulphide (MoS2), a compound more familiar to mechanical engineers as a motor lubricant. Researchers at EPFL in Switzerland found in 2011 that sheets of MoS2 can have very similar properties to those of graphene, but with one important advantage: MoS2 has a bandgap. A year later, a team from MIT that had also worked with graphene found MoS2 somewhat easier to work with and could create a number of useful components. The scientists found that it was possible to deposit layers of MoS2 on conventional wafers using the same kind of chemical vapour deposition processes as used for silicon epitaxy, although initial devices were made from flakes using a manufacturing technique similar to that used for experimental graphene circuits. As well as handling electronic circuits, as a direct-gap semiconductor MoS2 has potential applications in optoelectronics as a light generator or detector. What graphene and MoS2 do for electrons in n-type semiconductors, phosphorene could do for the holes in an analogous p-type semiconductor. The material, devised at Purdue University and unveiled earlier in 2014, is a 2D form of black phosphorus that provides very high mobility for holes, rather than electrons, as well as a bandgap. Although work is ongoing to understand how carriers move through the material, phosphorene has a heavily ridged structure that means carriers move across it in a number of different ways. Even silicon could see a late comeback in the era of 2D materials. Silicene was made in 2012 but, unlike the other materials, has strong interactions with its substrates, making it hard to exploit its native properties. However, if work is successful in isolating silicene electrically, the material may be a useful contender in the long term. One major problem with many 'graphene-alikes' is their reactivity. Devices will need to be carefully encapsulated to prevent them oxidising and losing their attractive properties. Although the 2D nature of 'graphene-alikes' fits well with conventional approaches to semiconductor design and manufacture, these devices cannot be made much smaller than the last generation of silicon CMOS transistors – we simply run out of atoms. One possibility is to move into the third dimension and harness the ability of another class of exotic materials, dubbed '3D graphene', to control electron behaviour in three spatial dimensions. The best example so far of this is trisodium bismuth, which exhibits a number of unusual quantum states, including those of topological insulators – which conduct along their surfaces, but not through their bulk – and Weyl semi-metals, where energy bands touch at various, regular points to form highly conductive paths within the material itself. In effect, Dirac cones are distributed throughout the material, turning a block of the material into a 3D analogue of graphene. One important characteristic of the material is that electrons preserve their spin states as they move, opening the door to spintronic devices similar to the tunnel junctions used in magnetic memories. Such devices would make it possible to combine memory and processing within components. Few expect trisodium bismuth to be viable for future electronic devices, but it is simply the best so far of the 3D topological Dirac semi-metals that we know about. Computational materials design techniques are already being used to find simpler 2D topological insulators. As materials scientists now have much better ideas of the quantum properties they want, computers can select candidate elements, simulate combinations of them and determine which of the thermodynamically stable alloys have the most promising properties. A crossover has already emerged between these materials and the chalcogenides proposed for phase change memories, which still remain something of a mystery in terms of their theoretical operation. Graphene could quickly move from being strong contender for silicon's replacement to being the prototype for more complex, unusual materials that help keep integration on track by pushing into the third dimension.