It is not easy finding a worthy successor to highly refined microchip technologies. But electronic devices built from molecular-scale components are fast becoming a good bet.
Unless you've been on a long vacation recently, you have probably heard a lot of hype about nanotechnology. Although many alternatives to existing microsystems are being considered, few would deny that there are formidable obstacles along this path. For example, the success of 'molecular electronics' — a potential successor to microelectronics — depends on the development of so-called bottom-up manufacturing techniques, in which molecular-scale components are chemically synthesized in large numbers and assembled into useful circuits1. The difficulties are daunting but, as Duan et al. show on page 66 of this issue2, real molecular-electronics devices are getting tantalizingly close. These authors have built nanometre-scale transistors and the smallest ever light-emitting diodes by exploiting two recent developments in the bottom-up approach to nanotechnology.
The first development was the perfection of techniques for growing nanometre-scale semiconductor wires. Until recently, such nanowires could be made only by depositing material on linear templates, such as natural fibres, the edges of thin films and the inside of membrane pores. The resulting wire structures are often polycrystalline and non-uniform. In contrast, the new methods require no template and produce nearly perfect single-crystal nanowires. The original idea goes back to the 1960s, when vapour-growth techniques were developed to produce crystalline semiconductors from hot gaseous reactants. It was discovered that whiskers of the semiconductor would spontaneously grow out of gold particles3 placed in the reaction chamber. This is because atoms in the gases, such as gallium and arsenide, dissolve in the molten metal much more easily than they attach to the surface of existing crystals. Once the metal is saturated, the atoms readily pass out of solution onto the end of a monocrystalline whisker growing out of the particle (Fig. 1a).
The latest techniques are clever variations on this theme, which use much smaller (nanometre scale) metal particles that may be sitting on a substrate, suspended in a colloidal solution4, or floating in the hot gas. Duan and others in Charles Lieber's group at Harvard5 have perfected the latter method to produce a wide variety of semiconductor wires, with well-controlled diameter, length and composition. Lieber's group have even shown that chemical dopants (impurities that add or remove electrons) can be incorporated during the growth of the nanowires, thereby controlling whether the nanowires are n-doped (having extra conduction electrons) or p-doped (with some electrons removed to leave positively charged 'holes').
The second development concerns an old method for positioning objects such as nanowires where they are wanted. Because a nanowire is elongated, and so easily electrically polarized, it is attracted towards a high electric field, with which it lines up. So when a voltage is applied between two electrodes, a nearby nanowire suspended in liquid is drawn in to bridge the gap between them (Fig. 1b). Recent work has produced orderly rows of parallel single-nanowire bridges in this way6. Duan et al.2 have now created a junction by placing a p-doped and an n-doped nanowire across each other (Fig. 1c).
Strong competitors for the same jobs as nanowires are single-walled carbon nanotubes, which are seamless hollow cylinders rather than solid rods. Both nanowires and nanotubes can be many micrometres long, making them far easier to work with than other popular molecular toys, such as conjugated molecules and nanocrystals, which are a thousand times shorter. This is why molecular field-effect transistors have previously been made only from nanotubes7, and now by Lieber's group from nanowires2. Unlike nanowires, nanotubes have completely smooth surfaces free from dangling bonds. They can be either semiconducting or highly metallic, and also have tiny diameters (down to less than 1 nm), giving them fascinating one-dimensional electronic properties8. But control of the diameter, crystal-lattice orientation and doping seems to be much easier in nanowires, and their extra rigidity can make it easier to assemble them into position. Another big advantage is that semiconductors are known to emit light, something that has never been seen in carbon nanotubes.
Junctions between crossed metallic and p-doped semiconducting nanotubes have previously been shown to behave as electronic diodes7 — that is, current flows when a positive voltage is applied to the p-doped tube, but there is little response when the voltage is negative. The crossed nanowires of Duan et al. also make diodes, but with a remarkable new feature: when a positive voltage is applied to the p-doped nanowire, light is emitted from the junction. The luminescence process is just the same as in regular, large light-emitting diodes (LEDs), in which electrons and holes injected across a p–n junction pair up and recombine to produce a photon. The authors also find that the frequency of the light increases if thinner nanowires are used, as the quantum states of the electrons inside them are squeezed into a smaller space, shifting their energy levels.
Assuming that their current-to-light efficiency can be improved — it is low because of non-luminescent processes that radiate heat rather than light — the nano-LEDs created by Duan et al. offer a range of exciting new possibilities in optoelectronics. The colour (frequency) of light they emit can be varied by using nanowires of different thicknesses, as well as by the usual method of using semiconductors that naturally luminesce at different frequencies. If the non-luminescent processes can be strongly suppressed, a single nanowire may become the first one-dimensional laser. Also, these are the first LEDs to reach sub-wavelength dimensions — that is, they are smaller than the wavelength of the light they emit. Finally, if this assembly technique can be scaled up to produce a grid of crossed nanowires, they would form an extraordinarily high-resolution LED array. Such a device might have a chance of being the first bottom-up electronics ever to see the light.
About this article
Applied Physics A (2011)