Light-Emitting Diode (LED)

Background

Light-emitting diodes (LEDs)—small colored lights available in any electronics store—are ubiquitous in modern society. They are the indicator lights on our stereos, automobile dashboards, and microwave ovens. Numeric displays on clock radios, digital watches, and calculators are composed of bars of LEDs. LEDs also find applications in telecommunications for short range optical signal transmission such as TV remote controls. They have even found their way into jewelry and clothing—witness sun visors with a series of blinking colored lights adorning the brim. The inventors of the LED had no idea of the revolutionary item they were creating. They were trying to make lasers, but on the way they discovered a substitute for the light bulb.

Light bulbs are really just wires attached to a source of energy. They emit light because the wire heats up and gives off some of its heat energy in the form of light. An LED, on the other hand, emits light by electronic excitation rather than heat generation. Diodes are electrical valves that allow electrical current to flow in only one direction, just as a one-way valve might in a water pipe. When the valve is "on," electrons move from a region of high electronic density to a region of low electronic density. This movement of electrons is accompanied by the emission of light. The more electrons that get passed across the boundary between layers, known as a junction, the brighter the light. This phenomenon, known as electroluminescence, was observed as early as 1907. Before working LEDs could be made, however, cleaner and more efficient materials had to be developed.

LEDs were developed during the post-World War II era; during the war there was a potent interest in materials for light and microwave detectors. A variety of semiconductor materials were developed during this research effort, and their light interaction properties were investigated in some detail. During the 1950s, it became clear that the same materials that were used to detect light could also be used to generate light. Researchers at AT&T Bell Laboratories were the first to exploit the light-generating properties of these new materials in the 1960s. The LED was a forerunner, and a fortuitous byproduct, of the laser development effort. The tiny colored lights held some interest for industry, because they had advantages over light bulbs of a similar size: LEDs use less power, have longer lifetimes, produce little heat, and emit colored light.

The first LEDs were not as reliable or as useful as those sold today. Frequently, they could only operate at the temperature of liquid nitrogen (-104 degrees Fahrenheit or -77 degrees Celsius) or below, and would burn out in only a few hours. They gobbled power because they were very inefficient, and they produced very little light. All of these problems can be attributed to a lack of reliable techniques for producing the appropriate materials in the 1950s and 1960s, and as a result the devices made from them were poor. When materials were improved, other advances in the technology followed: methods for connecting the devices electronically, enlarging the diodes, making them brighter, and generating more colors.

The advantages of the LED over the light bulb for applications requiring a small light source encouraged manufacturers like Texas Instruments and Hewlett Packard to pursue the commercial manufacture of LEDs. Sudden widespread market acceptance in the 1970s was the result of the reduction in production costs and also of clever marketing, which made products with LED displays (such as watches) seem "high tech" and, therefore, desirable. Manufacturers were able to produce many LEDs in a row to create a variety of displays for use on clocks, scientific instruments, and computer card readers. The technology is still developing today as manufacturers seek ways to make the devices more efficiently, less expensively, and in more colors.

Raw Materials

Diodes, in general, are made of very thin layers of semiconductor material; one layer will have an excess of electrons, while the next will have a deficit of electrons. This difference causes electrons to move from one layer to another, thereby generating light. Manufacturers can now make these layers as thin as .5 micron or less (1 micron = 1 ten-thousandth of an inch).

Impurities within the semiconductor are used to create the required electron density. A semiconductor is a crystalline material that conducts electricity only when there is a high density of impurities in it. The slice, or wafer, of semiconductor is a single uniform crystal, and the impurities are introduced later during the manufacturing process. Think of the wafer as a cake that is mixed and baked in a prescribed manner, and impurities as nuts suspended in the cake. The particular semiconductors used for LED manufacture are gallium arsenide (GaAs), gallium phosphide (GaP), or gallium arsenide phosphide (GaAsP). The different semiconductor materials (called substrates) and different impurities result in different colors of light from the LED.

Impurities, the nuts in the cake, are introduced later in the manufacturing process; unlike imperfections, they are introduced deliberately to make the LED function correctly. This process is called doping. The impurities commonly added are zinc or nitrogen, but silicon, germanium, and tellurium have also been used. As mentioned previously, they will cause the semiconductor to conduct electricity and will make the LED function as an electronic device. It is through the impurities that a layer with an excess or a deficit of electrons can be created.

To complete the device, it is necessary to bring electricity to it and from it. Thus, wires must be attached onto the substrate. These wires must stick well to the semiconductor and be strong enough to withstand subsequent processing such as soldering and heating. Gold and silver compounds are most commonly used for this purpose, because they form a chemical bond with the gallium at the surface of the wafer.

LEDs are encased in transparent plastic, rather like the lucite paperweights that have objects suspended in them. The plastic can be any of a number of varieties, and its exact optical properties will determine what the output of the LED looks like. Some plastics are diffusive, which means the light will scatter in many directions. Some are transparent, and can be shaped into lenses that will direct the light straight out from the LED in a narrow beam. The plastics can be tinted, which will change the color of the LED by allowing more or less of light of a particular color to pass through.

Design

Several features of the LED need to be considered in its design, since it is both an electronic and an optic device. Desirable optical properties such as color, brightness, and efficiency must be optimized without an unreasonable electrical or physical design. These properties are affected by the size of the diode, the exact semiconductor materials used to make it, the thickness of the diode layers, and the type and amount of impurities used to "dope" the semiconductor.

The Manufacturing
Process

Making semiconductor wafers

• 1 First, a semiconductor wafer is made. The particular material composition—GaAs, GaP, or something in between—is determined by the color of LED being fabricated. The crystalline semiconductor is grown in a high temperature, high pressure chamber. Gallium, arsenic, and/or phosphor are purified and mixed together in the chamber. The heat and pressure liquify and press the components together so that they are forced into a solution. To keep them from escaping into the pressurized gas in the chamber, they are often covered with a layer of liquid boron oxide, which seals them off so that they must "stick together." This is known as liquid encapsulation, or the Czochralski crystal growth method. After the elements are mixed in a uniform solution, a rod is dipped into the solution and pulled out slowly. The solution cools and crystallizes on the end of the rod as it is lifted out of the chamber, forming a long, cylindrical crystal ingot (or boule) of GaAs, GaP, or GaAsP. Think of this as baking the cake.
• 2 The boule is then sliced into very thin wafers of semiconductor, approximately 10 mils thick, or about as thick as a garbage bag. The wafers are polished until the surfaces are very smooth, so that they will readily accept more layers of semiconductor on their surface. The principle is similar to sanding a table before painting it. Each wafer should be a single crystal of material of uniform composition. Unfortunately, there will sometimes be imperfections in the crystals that make the LED function poorly. Think of imperfections as unmixed bits of flower or sugar suspended in the cake during baking. Imperfections can also result from the polishing process; such imperfections also degrade device performance. The more imperfections, the less the wafer behaves like a single crystal; without a regular crystalline structure, the material will not function as a semiconductor.
• 3 Next, the wafers are cleaned through a rigorous chemical and ultrasonic process using various solvents. This process removes dirt, dust, or organic matter that may have settled on the polished wafer surface. The cleaner the processing, the better the resulting LED will be.

Adding epitaxial layers

• 4 Additional layers of semiconductor crystal are grown on the surface of the wafer, like adding more layers to the cake. This is one way to add impurities, or dopants, to the crystal. The crystal layers are grown this time by a process called Liquid Phase Epitaxy (LPE). In this technique, epitaxial layers—semiconductor layers that have the same crystalline orientation as the substrate below—are deposited on a wafer while it is drawn under reservoirs of molten GaAsP. The reservoirs have appropriate dopants mixed through them. The wafer rests on a graphite slide, which is pushed through a channel under a container holding the molten liquid (or melt, as it is called). Different dopants can be added in sequential melts, or several in the same melt, creating layers of material with different electronic densities. The deposited layers will become a continuation of the wafer's crystal structure.

  LPE creates an exceptionally uniform layer of material, which makes it a preferred growth and doping technique. The layers formed are several microns thick.

• 5 After depositing epitaxial layers, it may be necessary to add additional dopants to alter the characteristics of the diode for color or efficiency. If additional doping is done, the wafer is again placed in a high temperature furnace tube, where it is immersed in a gaseous atmosphere containing the dopants—nitrogen or zinc ammonium are the most common. Nitrogen is often added to the top layer of the diode to make the light more yellow or green.

Adding metal contacts

• 6 Metal contacts are then defined on the wafer. The contact pattern is determined in the design stage and depends on whether the diodes are to be used singly or in combination. Contact patterns are reproduced in photoresist, a light-sensitive compound; the liquid resist is deposited in drops while the wafer spins, distributing it over the surface. The resist is hardened by a brief, low temperature baking (about 215 degrees Fahrenheit or 100 degrees Celsius). Next, the master pattern, or mask, is duplicated on the photoresist by placing it over the wafer and exposing the resist with ultraviolet light (the same way a photograph is made from a negative). Exposed areas of the resist are washed away with developer, and unexposed areas remain, covering the semiconductor layers.
• 7 Contact metal is now evaporated onto the pattern, filling in the exposed areas. Evaporation takes place in another high temperature chamber, this time vacuum sealed. A chunk of metal is heated to temperatures that cause it to vaporize. It condenses and sticks to the exposed semiconductor wafer, much like steam will fog a cold window. The photoresist can then be washed away with acetone, leaving only the metal contacts behind. Depending on the final mounting scheme for the LED, an additional layer of metal may be evaporated on the back side of the wafer. Any deposited metal must undergo an annealing process, in which the wafer is heated to several hundred degrees and allowed to remain in a furnace (with an inert atmosphere of hydrogen or nitrogen flowing through it) for periods up to several hours. During this time, the metal and the semiconductor bond together chemically so the contacts don't flake off.
• 8 A single 2 inch-diameter wafer produced in this manner will have the same pattern repeated up to 6000 times on it; this gives an indication of the size of the finished diodes. The diodes are cut apart either by cleaving (snapping the wafer along a crystal plane) or by sawing with a diamond saw. Each small segment cut from the wafer is called a die. A difficult and error prone process, cutting results in far less than 6000 total useable LEDs and is one of the biggest challenges in limiting production costs of semiconductor devices.

Mounting and packaging

• 9 Individual dies are mounted on the appropriate package. If the diode will be used by itself as an indicator light or for jewelry, for example, it is mounted on two metal leads about two inches long. Usually, in this case, the back of the wafer is coated with metal and forms an electrical contact with the lead it rests on. A tiny gold wire is soldered to the other lead and wire-bonded to the patterned contacts on the surface of the die. In wire bonding, the end of the wire is pressed down on the contact metal with a very fine needle. The gold is soft enough to deform and stick to a like metal surface.
• 10 Finally, the entire assembly is sealed in plastic. The wires and die are suspended inside a mold that is shaped according to the optical requirements of the package (with a lens or connector at the end), and the mold is filled with liquid plastic or epoxy. The epoxy is cured, and the package is complete.

Quality Control

Quality in semiconductor manufacturing takes two forms. The first concern is with the final produced product, and the second with the manufacturing facility. Every LED is checked when it is wire bonded for operation characteristics. Specific levels of current should produce specific brightness. Exact light color is tested for each batch of wafers, and some LEDs will be pulled for stress testing, including lifetime tests, heat and power breakdown, and mechanical damage.

In order to produce products consistently, the manufacturing line has to operate reliably and safely. Many of the processing steps above can be automated, but not all are. The general cleanliness of the facility and incoming blank wafers is monitored closely. Special facilities ("clean rooms") are built that keep the air pure up to one part in 10,000 for particular processing steps (particularly numbers 1-5 above). All of these checks arise from a desire to improve the yield, or the number of successful LEDs per wafer.

The Future

Optoelectronics is blossoming with the advent of better and better processing techniques. It is now possible to make wafers with a purity and uniformity unheard of 5 years ago. This will effect how bright and how efficient LEDs can be made, and how long they will last. As they get better, they are appropriate for increasingly demanding applications, such as communications. The average lifetime of a small light bulb is 5-10 years, but the average modern LED should last 100 years before failure. This makes them suitable for applications where it is difficult or impossible to replace parts, such as undersea or outerspace electronics. Although LEDs are inappropriate for long-range optical fiber transmission, they are often useful for short range optical transmission such as remote controls, chip to chip communication, or excitation of optical amplifiers.

Other materials are being developed that will allow fabrication of blue and white light LEDs. In addition to making possible a wider variety of indicators and toys with more colors, blue light is preferable for some applications such as optical storage and visual displays. Blue and white light are easier on the eyes. Additional colors would certainly open up new applications.

Finally, as process technology advances and it becomes possible to incorporate more devices on a single chip, LED displays will become more "intelligent." A single microchip will hold all the electronics to create an alphanumeric display, and will make instrumentation smaller and more sophisticated.

Where To Learn More

Books

Bergh, A. A. and P. J Dean. Light-Emitting Diodes. Clarendon Press, 1976.

Gillessen, Klaus. Light-Emitting Diodes: An Introduction. Prentice Hall, 1987.

Optoelectronics/Fiber-Optics Applications Manual. McGraw-Hill, 1981.

Understanding Solid State Electronics. Radio Shack/Texas Instruments Learning Center, 1978.

Williams, E. W. and R. Hall. Luminescence and the Light-Emitting Diode. Pergamon Press, 1978.

Periodicals

Cole, Bernard C. "Now a LED Can Take On the Light Bulb." Electronics. October, 1988, p. 41.

Iversen, Wesley R. "Would You Believe LED Brake Lights." Electronics. September 18, 1986.

Marston, Ray. "Working with LED's." Radio-Electronics. January, 1992, p. 50; February, 1992, p. 69.

Weisburd, Stefi. "Silicon Devices: LED There Be Light." Science News. May 9, 1987, p. 294.

—Leslie G. Melcer

Semiconductors

The development of semiconductors is clearly among the most significant technological achievements to evolve from the study of solid-state chemistry and physics. Aside from their well-known applications in computers and electronics, semiconductors are also used in a wide variety of optical devices such as lasers, light-emitting diodes, and solar panels. The diversity of applications can be readily understood with only a basic understanding of the theory behind these materials.

Theory

The operation of semiconductors is best understood using band theory. At its most fundamental level, band theory can be extremely complex, requiring relatively advanced mathematics and physics. When a large number of atoms combine to form a solid, the electrons e ⁻ in the solid are distributed into energy bands among all the atoms in the solid. Each band has a different energy, and the electrons fill these bands from the lowest energy to the highest, similar to the way electrons occupy the orbitals in a single atom. The variation in properties between electrical insulators, conductors (metals ), and semiconductors stems from differences in the band structures of these materials (see Figure 1). For this discussion, three terms must be defined. The highest energy band that contains electrons is called the valence band, whereas the lowest energy empty band is called the conduction band. The band gap is the difference in energy between the valence and conduction bands. The laws of quantum mechanics forbid electrons from being in the band gap; thus, an electron must always be in one of the bands.

In a metal (e.g., copper or silver), the valence band is only partially filled with electrons (Figure 1a). This means that the electrons can access empty areas within the valence band, and move freely across all atoms that make up the solid. A current can therefore be generated when a voltage is applied. In general, for electrons to flow in a solid, they must be in a partially filled band or have access to a nearby empty band. In an electrical insulator, there is no possibility for electron flow (Figure 1b), because the valence band is completely filled with electrons, and the conduction band is too far away in energy to be accessed by these electrons (the band gap is too large). A semiconductor (Figure 1c) is a special case in which the band gap is small enough that electrons in the valence band can jump into the conduction band using thermal energy. That is, heat in the material

(even at room temperature) gives some of the electrons enough energy to travel across the band gap. Thus, an important property of semiconductors is that their conductivity increases as they are heated up and more electrons fill the conduction band. The most well-known semiconductor is silicon (Si), although germanium (Ge) and gallium arsenide (GaAs) are also common.

To complete the development of semiconductor theory, the concept of doping must be described (see Figure 2). In principle, the idea is to introduce a different kind of atom into a semiconductor in order to modify its electronic structure. Consider, for example, adding a small amount of phosphorus, P, into a silicon host. Phosphorus is one column to the right of silicon in the Periodic Table, so it contains one additional electron. This means that doping P into Si has the effect of introducing additional electrons to the material, such that some e ⁻ must go into the conduction band. Because extra negatively charged electrons are added to the system, phosphorusdoped Si is called an n- type semiconductor, and phosphorus is described as a donor (of electrons). Similarly, a p- type semiconductor can be fabricated by adding an element to the left of Si in the Periodic Table. Boron, B, is a common dopant for a p- type. In this case, the valence band will be missing electrons. These empty locations in a p- type semiconductor are also referred to as holes. Since holes represent the absence of an electron, they carry a positive charge. In p- type semiconductors, boron is referred to as an acceptor (of electrons). From Figure 2, it can be seen that both n- and p- type materials create partially filled bands, allowing for electrical conduction. Dopant concentrations are fairly small, around 10¹⁶ atoms/cm³, constituting only about ten-billionths of the total mass of the material.

If p- and n- type materials are layered together, a p-n junction results (Figure 2c). Right at the interface, some of the excess electrons from the n- type combine with holes from the p- type. The resulting charge separation creates an energy barrier that impedes any further movement of electrons. In most technological applications, the important properties of semiconductors are the result of the band structure of the p-n junction. A single

junction based on the same host material (e.g., one interface of p- and n- doped silicon) is called a homojunction. The homojunction model is used here to describe the properties of many devices that are based on semiconductors. However, it should be noted that real systems are typically composed of multiple p-p, n-n, and p-n junctions, called heterojunctions. Such configurations greatly improve the performance of these materials; in fact, the development of heterojunction devices was critical to the widespread practical application of this technology.

Semiconductors in Electronics

Semiconductors are used extensively in solid-state electronic devices and computers. The majority of materials for these applications are based on doped silicon. An important property of p-n junctions is that they allow electron flow only from the n side to the p side. Such one-way devices are called diodes. Consider Figure 2c again. If a positive voltage (also called a forward bias) is applied that lowers the energy barrier between n and p, then the electrons in the conduction band on the n side can flow across the junction (and holes can flow from p to n ). A reverse bias, however, raises the height of the barrier and increases the charge separation at the junction, impeding any flow of electrons from p to n.

Diodes have several important applications in electronics. The power supplied by most electrical utilities is typically alternating current (AC); that is, the direction of current flow switches back and forth with a frequency of sixty cycles per second. However, many electronic devices require a steady flow of current in one direction (direct current or DC). Since a diode only allows current to flow through it in one direction, it can be combined with a capacitor to convert AC input to DC output. For half the AC cycle, the diode passes current and the capacitor is charged up. During the other half of the cycle, the diode blocks any current from the line, but current is provided to the circuit by the capacitor. Diodes applied in this way are referred to as rectifiers.

The by far most important application of semiconductors is as logic gates and transistors in computers. Logic gates, such as OR and AND gates, take advantage of the one-way nature of diodes to compare the presence or

absence of current at different locations in a circuit. More complex solid-state transistors are composed of npn or pnp junctions. The device geometry is slightly more complicated than that observed in a diode, but the result is materials that allow for the generation of the zeros and ones required for the binary logic used by computers.

Optoelectronic Devices

Optoelectronic materials are a special class of semiconductors that can either convert electrical energy into light or absorb light and convert it into electrical energy. Light-emitting diodes (LEDs), for example, are commonly used for information display and in automotive interior lighting applications. In an LED, a forward bias applied across the junction moves electrons in the conduction band over holes in the valence band. The electron and hole combine at the junction, and the energy created by this process is conserved via the emission of light (Figure 3a). The wavelength of emitted light will depend on the band gap of the material; larger band gaps lead to shorter wavelengths of light. Only certain kinds of semiconductors, called direct gap semiconductors, exhibit this behavior. GaAs is an example of a direct gap semiconductor used in these applications. Silicon is an indirect gap material, and electrons and holes combine with the generation of heat instead of light.

A diode laser operates in essentially the same fashion as an LED. Two additional requirements must be met for a direct gap semiconductor to be an efficient laser. The first is that larger forward bias currents are needed for a laser than for an LED, because lasers require a higher degree of population inversion—a large number of electrons in the conduction band above empty levels in the valence band. Lasers also require an optical cavity; light bounces back and forth within the cavity, building up intensity. In a diode

laser, this can be achieved by cleaving and polishing opposite faces of the diode. The smooth faces act like partially reflecting mirrors. This kind of laser is used to read information on compact disks and is also used in laser pointers.

The most common materials for lasers and LEDs are heterojunctions based on GaAs. More complex systems containing Ga, As, P, Al, and N are also used. The band gap of these materials can be tailored to create emission from infrared to yellow. In optical data storage systems, such as compact disks, the amount of information that can be stored is dependent in part on the wavelength of light being used to read the disk—shorter wavelengths allow for denser information storage. Thus, there has been considerable interest in developing larger band gap LEDs and lasers that emit in the blue. This has been achieved in semiconductors based on GaN (gallium nitride). Further refinement of these materials will no doubt lead to significant advances in optoelectronic technology in the coming years.

A final important class of optoelectronic devices based on semiconductors is photovoltaics, such as photodetectors and solar cells. In some respects, these can be regarded as LEDs operating in reverse. Light energy incident on the p-n junction is absorbed by an electron, which then jumps to the conduction band (Figure 3b). Once in the conduction band, the electron travels downhill (energetically) to the n side of the junction, with a hole migrating to the p side. This creates a flow of current that is the reverse of what is seen in a forward biased diode. The result is the conversion of light energy to electrical energy. These devices can therefore be used to detect light, as in digital imaging systems or miniature cameras; or the electrical energy can be stored, as in solar cells. Commercial photovoltaics are based on a variety of host materials, including Si, AlGaAs, and InAlAs.

Fabrication

The industrial fabrication of semiconductors can be extremely complex, involving high-purity materials, sophisticated equipment, and hundreds of steps. Most processes begin with the growth of a large single crystal of n- type Si, called a wafer. A dopant (e.g., phosphorus) is added to high-purity molten silicon, and a crystal is then slowly extracted from this melt. The polished wafer is 20 to 30 centimeters (7.9–11.8 inches) in diameter.

The rest of the processing will depend on the nature of the device being produced. A simple p-n junction is usually fabricated via photolithography and etching processes. In this method, a layer of silicon dioxide, SiO₂, is created on the surface of the wafer by heating it in the presence of

oxygen. Some of the SiO₂ is then chemically stripped away, or etched, exposing only a portion of the Si wafer. This exposed part of the wafer is made into p- type material by bombarding it with boron ions. As these ions diffuse into the Si wafer, p- type Si is formed. Since the original wafer was n- type, a p-n junction forms where the diffusion of boron stops. Metal contacts can then be added to each side of the junction to create a simple homojunction device.

Fabrication of more complicated devices is achieved via combinations of etching, deposition, and ion implantation steps. In the production of integrated circuits for computers, about 400 chips can be synthesized on a single 30-centimeter (11.8-inch) wafer. Each chip may contain as many as 50 million transistors in a space barely more than 1 centimeter (0.39 inches) on a side—a truly remarkable technological achievement. As faster and faster systems are developed, the demand for smaller and smaller features increases. Such miniaturization is the most significant challenge facing the semiconductor industry today.

Semiconductors are used in a wide variety of electronic and optoelectronic applications. The useful properties of semiconductors arise from the unique behavior of doped materials, the special control of electron flow provided by p-n junctions, and the interaction of light energy with electrons at these junctions. The industry continues to grow, and research in this and related areas (i.e., organic semiconductors and molecular transistors) is occurring at academic institutions around the world.

see also Germanium; Silicon.

Anthony Diaz

Bibliography

Bhattacharya, Pallab (1994). Semiconductor Optoelectronic Devices. Englewood Cliffs, NJ: Prentice Hall.

May, Gary S., and Sze, Simon M. (2003). Fundamentals of Semiconductor Fabrication. New York: Wiley.

Miessler, Gary L., and Tarr, Donald A. (1999). Inorganic Chemistry, 2nd edition, Upper Saddle River, NJ: Prentice Hall.

Myers, H. P. (1997). Introductory Solid State Physics, 2nd edition. Philadelphia: Taylor & Francis.

Svelto, Orazio (1989). Principles of Lasers, 3rd edition, tr. and ed. David C. Hanna. New York: Plenum.

Texas Instruments Learning Center (1972). Understanding Solid State Electronics, 2nd edition. Dallas: Texas Instruments Inc.

Wold, Aaron, and Dwight, Kirby (1993). Solid State Chemistry. New York: Chapman & Hall.

SEMICONDUCTORS

SEMICONDUCTORS are solid materials with a level of electrical conductivity between that of insulators and conductors. Beginning with semiconducting elements found in nature, such as silicon or germanium, scientists learned to enhance and manipulate conductivity by changing the configuration of electrons in the material through the combination of materials and the precise introduction of impurities. Because of their ability to control electrical currents, semiconductors have been used in the manufacture of a wide range of electronic devices—including computers—that changed American life during the second half of the twentieth century.

Although the scientific study of semiconductors began in the nineteenth century, concentrated investigation of their use did not begin until the 1930s. The development of quantum physics during the first third of the twentieth century gave scientists the theoretical tools necessary to understand the behavior of atoms in solids, including semiconductors. But it was a commercial need that really stimulated semiconductor research in the United States. The rapid growth of the national telephone network had by 1930 made the replacement of mechanical switches—too large and too slow for the expanding system—highly desirable. Vacuum tubes, used in radios and other devices, were too expensive and fragile for use in the telephone network, so researchers turned their focus to solid crystals. Radar research during World War II, through efforts to make reliable and sensitive transmitters and receivers, advanced understanding of the relative merits of different crystal substances. Germanium and silicon showed the most promise. Scientists at Bell Laboratories, the research arm of the AT&T Corporation, built upon wartime investigations done there and elsewhere to design the first transistor using the semiconductor germanium. A prototype was produced in 1947, and innovation followed rapidly. William Shockley, Walter Brattain, and John Bardeen, all Bell Labs researchers, were awarded the 1956 Nobel Prize in physics for their research on semi-conductors and the design of the transistor.

Transistors replaced vacuum tubes in electronic devices slowly at first. Hearingaids were the first technology to use the new, small transistors, but it was inexpensive portable radios that created the first large commercial market for the device. Initially manufactured by the Texas Instruments Company, the first large semiconductor company, "transistor" radios soon became a specialty of manufacturers in the Far East. Not limiting themselves to these consumer products and military signal devices, American researchers and manufacturers sought ways to use germanium-based transistors in computing machines. The more versatile silicon, however, ultimately replaced germanium to satisfy the needs of evolving computer technology.

The semiconductor silicon gave its name to a region—an area between San Jose and San Francisco, California, that became known as Silicon Valley—and fomented revolutions in technology, business, and culture. Silicon Valley grew outward from Palo Alto, home to Stanford University and host to a number of electronic pioneers beginning in the 1920s with the vacuum tube researcher Lee De Forest. Once scientists had determined that silicon had the necessary properties for applications in computing, practical concerns took center stage. Although silicon is one of the most common elements on earth—sand is made of silicon and oxygen—isolating and purifying it is notoriously difficult. But interest in silicon-based devices was very strong, and by the late 1950s a diversified semiconductor industry was developing, centered in California but serving government and commercial clients throughout the country.

The electronics industry initially turned to semiconducting materials to replace large, slow, electromechanical switches and fragile, unreliable vacuum tubes. But the new technology proved to be far more than an incremental improvement. Semiconductors showed promise for miniaturization and acceleration that previously seemed fanciful. An insatiable desire for faster, smaller devices became the driving force for the semiconductor industry. An impressive stream of innovations in theory, design, and manufacturing led the semiconductor industry to make ever-smaller, ever-faster devices for the next half century. Improvements in semiconductor devices led to faster, cheaper electronics of all kinds, and to the spread of the semiconductor and its dependent industries throughout the world.

Although the semiconductor industry was born and developed in the United States, manufacturing of silicon-based devices—including the "memory chips" that are most essential to computer and other electronic technologies—began to move overseas in the 1970s. Japan was a particularly strong participant in the manufacture of high-quality chips. While American companies were eager to buy from Japanese manufacturers, American semiconductor manufacturers turned to the government for support and market intervention. Struggles in the industry continued throughout the 1970s and 1980s but the great expansion of the market for computers and continuing innovation kept semiconductor-based businesses flourishing both in the United States and abroad.

Silicon, the premier semiconductor, belongs among a small number of other substances that have changed the course of history. Unlike earlier, comparably influential materials—salt and gold, for example—mastering the use of silicon required an enormous amount of research. In fact, silicon is the most studied substance in history. Semi-conductor science, and the industry it spawned, drew upon uniquely American elements in their development. Industrial research labs such as those at AT&T, IBM, and the entrepreneurial companies of Silicon Valley were vital to the development of the semiconductor industry, as was the government support of research during and after World War II. The military also influenced development as an important customer to the industry. The future will likely bring a replacement for silicon in the ongoing search for smaller, faster electronic devices, but silicon has earned a most valuable place in the history of technology and twentieth-century culture.

BIBLIOGRAPHY

Bassett, Ross. To the Digital Age: Research Labs, Startup Companies, and the Rise of MOS Technology. Baltimore: Johns Hopkins University Press, 2002.

Misa, Thomas J. "Military Needs, Commercial Realities, and the Development of the Transistor." In Military Enterprise and Technological Change: Perspectives on the American Experience. Edited by Merritt Roe Smith. Cambridge, Mass.: MIT Press, 1987.

Queisser, Hans. The Conquest of the Microchip. Cambridge, Mass.: Harvard University Press, 1990.

Loren ButlerFeffer

light-emit·ting di·ode • n. see LED.

light-emitting diode (LED) See LED display.