The next time you are behind the wheel of your car and a traffic light causes you to slam on brakes and 'see red', dissipate some of the incipient road-rage by taking a few seconds to examine the offending signal itself. More often than not, you'll be able to make out the filament of a single light bulb through the red plastic lens. Increasingly, however, especially if the signal is in an inaccessible place (such as high above the roadway) you'll be staring at a close-packed matrix of tiny lights: an array of light-emitting diodes. These unassuming elements have heralded a quiet revolution in consumer electronics; one that promises cheaper, brighter and more reliable lighting and display technology and which, in its own way, is as revolutionary as the invention of the integrated circuit. Increasingly ubiquitous, efficient, colourful and bright, they have found their way into digital watches, bicycle lamps, the blindingly bright sensors on the bottom of the new generation of optical computer mice, 'everlasting' torches and even the lighting of our homes.
How can one convert electricity into light? There are several ways. The first and least efficient way has been around with us since before even the days of Joseph Swan and Thomas Edison: use it to make a substance very hot. In its modern manifestation1, one takes a very thin tungsten filament, coils it up and then coils the coil again, puts it in an insulating argon atmosphere and passes an electric current through it. Electric current consists of the movement of electrons, and their motive energy gets converted into heat as they bash into the tungsten atoms. The tungsten gets hot, very hot, and the only way the energy can escape is by being converted into electromagnetic radiation: light - and heat to anyone who stands close enough.
This process is simple but fairly indiscriminate: lots of wavelengths of light get generated, and most of it goes to waste in the form of heat. If one wants to generate a specific colour of light, one has to winnow out the unwanted wavelengths by interposing a coloured filter between the observer and the filament, causing even more loss of light intensity. All these losses add up, and what is more, the light source itself is very prone to failure (this all-too-common event in itself being directly responsible for a gamut of execrable jokes).
There are better ways; the fluorescent lamp uses the motion of electrons in a gas to excite mercury atoms. These in turn radiate ultraviolet light that is converted to visible light by a phosphor coating on the inside of the lamp. They are more efficient than incandescent bulbs, last longer, but are more difficult to make and hence more expensive. They also take up a lot of space and look horrible into the bargain. Gas discharge lamps, the kind that power those bluish, very bright car headlamps one sometimes sees, function the same way but dispense with the phosphor, instead exciting the gas atoms into a wider range of energy states, many in the visible region of the spectrum. All these methods of light generation share a common drawback: they waste energy. They also tend to give rise to rather bulky and fragile light sources, which often require thousands of volts to make them work.
Light-emitting diodes dispense with the glass and the gas. Instead, a tiny chip of semiconductor sits inside a robust plastic housing and is connected to a battery. An equally tiny reflector funnels the light generated by the chip out of the front of the housing through a plastic lens. So how does the electric current get converted to light?
Semiconductors are so called because their electrical conductivity sits somewhere between that of a metal and an insulator. Electrons in a metal behave rather like balls on a snooker table: they can move around easily and only lose energy when they bang into each other or stationary obstacles. Metals therefore carry electric currents easily.
Electrons in an insulator behave like oranges stacked in a crate: they are so tightly wedged in together that they have no room to move. Those in a semiconductor are similar, but there is a vital difference. Imagine there is a second box, a foot or two above the first one, and that someone takes a few oranges from the lower box and puts them in the top one. Now not only can these liberated oranges move, but all the oranges in the lower box can as they shuffle around, each occupying the 'holes' left behind and creating new ones in their wake.
In a semiconductor, the oranges are electrons, and the two boxes are really energy 'bands', one separated from the other by a gap of several electron volts (the amount of energy an electron gains in travelling through a potential difference of one volt), but the analogy still holds. To get a semiconductor to conduct better, one has to move wedged-in electrons from the lower of the two bands to the higher by pumping in energy, either through heat or light. The 'holes' in the lower of the two bands behave like positive charges and so can also move under the influence of an electric field.
There is another way of making a semiconductor into a three-quarter- or seven-eighths-conductor: doping. Doping involves the introduction of tiny amounts of impurity atoms, generally phosphorus or gallium, which either inject holes or free electrons. The conductivity shoots up by several orders of magnitude when even tiny amounts of impurity are introduced. Semiconductors with surplus electrons are referred to as n-type (negative-type), whereas those with surplus holes are p-type (positive-type), after the polarity of the predominant charge carriers.
Diodes are the simplest solid-state electronic device: the equivalent of an electrical one-way valve. In a diode, n-type and p-type semiconductors are joined together, and the interface takes on some quite interesting properties. When a negative voltage is applied to the n-type and a positive to the p-type layers a current flows freely. Reverse the voltage, and the diode becomes as conducting as a house brick. The reverse voltage sucks holes and electrons out of the two layers of the diode and the device, having nothing to carry a current with, becomes insulating. In the conducting configuration, electrons and holes recombine at the interface between the two layers and emit energy, and are replenished continually from the external circuit.
Harvesting the Energy of Electrons
At an atomic level, the junction in a 'forward-biased' diode resembles a waterfall as electrons plummet from a high-energy state into holes that congregate at the bottom. The trick is to get the electrons to give up this energy in a productive fashion, rather like getting the waterfall to drive a waterwheel. This is rather more difficult than it sounds, due to some annoying fine print written into the laws of physics.
There are essentially two kinds of semiconductor: direct and indirect gap semiconductors. The former allow electrons to move from the higher to the lower band without having to change their speed. The latter require that not only does the electron lose energy, but also it slows down substantially before it switches bands. The electron has to lose energy by colliding with the crystal lattice, so the diode heats up, and most of the energy is wasted. Silicon is an indirect-gap semiconductor.
To get light out of a diode, one has to use a direct gap semiconductor with a sizable band gap. The first light emitting diode should really have been called an 'infra-red emitting diode' as it didn't produce anything visible. It was made out of a crystalline material called gallium arsenide; belonging to a class of semiconductor called the III-V semiconductors (after the numbers of the groups in the Periodic Table to which gallium and arsenic belong), and emitted near-infrared radiation.
In the late 1960s, more LEDs started to appear, this time capable of being seen. Typically they were made of gallium arsenide/gallium arsenide phosphide layered structures. These emitted a dull-red colour. Then orange, yellow and 'green'2 LEDs soon followed. As manufacturers perfected their processes, the quality and intensity of the light increased. Soon, LEDs began to find other applications, such as the now commonplace seven-segment number display, indicator lamps, and the signalling elements of infra-red remote controls.
Typically, LEDs will emit coloured light in a narrow band of energies centred around that of the band-gap. This implies that providing you are prepared to live with the limited colour range, the LED is much more efficient than a gas-discharge or incandescent lamp. An LED will have about 10% of the power consumption of an incandescent lamp of comparable brightness. It will also last for in excess of 10,000 hours before it needs to be replaced. 20% of the energy requirement of a developed country will be accounted for in the lighting of homes, offices and factories. Replace light bulbs by LEDs and one has already made a sizable impact on the greenhouse effect, not to mention the energy savings made through not having to manufacture replacements as frequently.
Miniaturising the Rainbow
Colour, however, depends entirely upon the gap between the two energy bands. This is an intrinsic property of the semiconductor used to make the LED, and not susceptible to much tinkering. The energy gap needs to be equal to or greater than the energy of the photons emitted: bluer colours require higher energies and hence a bigger band gap. For blue3 light, an energy gap of at least 2.5 electron volts is required. The only easily processable semiconductor with that size of gap was silicon carbide, more commonly used for tipping masonry drills and making sandpaper, and this was an indirect-gap semiconductor: blue light came out, but in a trickle rather than a flood. Display technologists who envisaged huge television screens made up of three colours of LED had to remain disappointed. Even those who wanted to make traffic lights from these had to make do with the approximate green of gallium phosphide LEDs.
Attempts were made to use the II-VI semiconductors: zinc sulphide, zinc selenide and suchlike in LEDs. These emitted blue light but not for very long: they were riddled with impurities and imperfections, and generally stopped working after a few hours. It wasn't until a few years ago that the physicists cracked the problem. A III-V material, gallium nitride (GaN), had been shown to emit blue light but was abandoned because there were problems in fabricating devices from it. A dogged and innovative Japanese scientist, Shuji Nakamura, solved this and other problems with GaN over a period of several years. Now it is possible to buy blue-green, blue4 and even ultraviolet LED lamps. Nichia Corporation5of Japan, one of the pioneers of the blue/green LED, now produce a true green LED that is so bright that it cannot be viewed for more than a few seconds without eye damage occurring. Blue LEDs are likely to come down in price substantially, as it has been very recently shown that GaN can be grown on cheap silicon substrates.
Even very bright white LEDs can now be bought: these work by coupling a blue LED with a phosphor, or even another semiconductor, which absorbs some of the blue light and emits it as yellow light. The combined mixture of blue and yellow light gives the impression of a bright white light. These are increasingly finding their way into hand torches due to their attractive combination of low power consumption and long life.
Pensioning-off the Cathode Ray Tube
The Cathode Ray Tube (CRT) has been the lynchpin of display technology ever since the invention of the iconoscope6. Little has changed between 1932 and the present day: only the invention of liquid crystal monitors has had any impact upon the way we watch television or interact with our computers.
CRTs create images by directing a beam of electrons at a phosphor-coated tube. They are heavy, difficult to make, bad for the environment (all that lead glass needs to disposed-of carefully!), fragile, and have a nasty habit of imploding once their screen size exceeds 40 inches. Liquid crystal monitors are slow to respond, expensive, even more fragile and not particularly bright. If a viewer wants a wide screen television then they either have to put up with a big, cumbersome conventional telly, or dig deep into their pockets for a liquid crystal screen.
Where LEDs could help resolve this dilemma is by providing compact, cheap, bright polychromatic light sources. It's a well-known fact that combining various ratios of the three primary colours - red, blue and green7 - can simulate any colour. A display screen made up of equal numbers of tiny coloured LEDs should be able to reproduce any image clearly and brightly.
This is fine in theory, but to achieve this using present day technology, one has to assemble huge numbers of LEDs together to get an image of sufficient resolution. This has been done with large display boards: the manufacturer simply buys tens of thousands of the LED lamps and wires them in place. For a computer monitor, where each picture element is about the size of a pinprick, this approach is not an option.
However, at this point the chemist, like a knight in a shining white lab coat, again rides to the rescue of the physicist. Informed tinkering can control virtually all properties of an organic compound. If one can make an organic chemical that can convert electricity to light, there's a good chance that the colour of it will be controllable. Many of these materials can be doped to form n-type or p-type semiconductors. Sandwiching layers of these materials together produces to organic LEDs, much like their inorganic counterparts.
What is more, organic chemicals tend to be a lot easier to process than inorganic ones. They often dissolve or melt easily. Sometimes they form resilient, flexible polymers. Specially adapted inkjet printers can spray patterns of these materials onto substrates to form compact, high-resolution display screens. Light-emitting polymers now provide the backlights of mobile phone displays.
Although this technology is still in its infancy, it promises cheaper, lighter, brighter and thinner television and computer screens. It may even give rise to 'electric wallpaper' that illuminates our homes evenly and cheaply.
The days of the CRT might not yet be over but, thanks to the humble LED, they are certainly numbered. And with any luck, we may eventually say goodbye once and for all to even the clunky incandescent bulb and its reviled cousin, the fluorescent tube. The future is bright, and it's not just orange.