Semiconductor Electronics

Semiconductor electronics is the branch of electrical engineering that makes devices out of materials whose ability to conduct electricity sits between that of a metal and an insulator—and, crucially, can be controlled.

10 frameworks, from 1931 to today · Electrical Engineering · Engineering

still a working view ended · each leads to its entry below

Overview

Semiconductor electronics is the branch of electrical engineering that makes devices out of materials whose ability to conduct electricity sits between that of a metal and an insulator—and, crucially, can be controlled12. A silicon crystal with a few parts per million of boron in it conducts differently from a pure one; apply a voltage across a junction of two such regions and current flows in one direction only; shine light on it and it produces a voltage; pass current through it and it emits light. Everything from a diode to a microprocessor rests on these controllable properties. What is at stake is the physical layer of modern technology: the transistor is the most manufactured object in history, and the field's central achievement has been to make these devices smaller, faster, cheaper, and more reliable, decade after decade, by understanding and manipulating the physics of solids.

From crystal detectors to the quantum theory of solids

The field's prehistory is the "cat's whisker" detector1. Around the turn of the twentieth century, radio pioneers found that a fine metal point pressed against certain crystals—galena, silicon carbide, and later silicon—would rectify a radio signal, passing current one way but not the other. These detectors worked, but erratically and mysteriously; nobody could say why a particular contact worked or failed1. The vacuum tube, which could be engineered predictably, displaced them for most purposes1.

The explanation came from solid-state physics. In the 1920s and 1930s, physicists developed the band theory of solids: in a crystal, electrons may occupy only certain ranges of energy—"bands"—separated by forbidden "gaps." A metal has a partly filled band, so electrons move freely. An insulator has a full band separated from the next empty one by a wide gap3. A semiconductor has a narrow gap, so a few electrons can be thermally excited across it, and the vacancies they leave behind—"holes"—behave as mobile positive charges. In 1931 Alan Wilson used band theory to explain why some materials are semiconductors and why their conductivity rises with temperature, unlike metals. The theory was qualitative but decisive: it turned a collection of puzzling materials into a class with a mechanism.

Doping, junctions, and the point-contact transistor

The next step was control. In the early 1940s, work at Purdue and elsewhere established that tiny impurities dominate a semiconductor's behavior. A "donor" impurity contributes an extra electron (n-type material); an "acceptor" creates a hole (p-type)1. This is doping, and it is the field's foundational act: a semiconductor's properties are engineered by what you put into it, at concentrations of parts per million.

The p-n junction followed14. Where n-type and p-type regions meet, electrons and holes diffuse into each other and leave behind fixed charged ions, creating a built-in electric field and a depletion region45. The junction rectifies: forward bias shrinks the barrier and current flows; reverse bias widens it and almost nothing passes1. In 1949 William Shockley published the junction theory that made this quantitative, and the p-n junction became the field's basic building block—the diode, and later the heart of the transistor, the solar cell, and the light-emitting diode.

The transistor itself arrived at Bell Telephone Laboratories in December 1947, when John Bardeen and Walter Brattain demonstrated a point-contact device: two metal contacts pressed onto a germanium surface, with the current through one controlled by the other167. It was a triumph of empirical tinkering guided by the surface physics of the day1. Shockley, dissatisfied with the point-contact design's fragility, pursued a different structure and in 1948 conceived the junction transistor—a sandwich of n-p-n or p-n-p material in which a thin base region controls current between emitter and collector167. The first junction transistors were made in 1950–51. The bipolar junction transistor (BJT) was manufacturable, predictable, and became the workhorse of the 1950s6. Its design—base width, doping profiles, and the trade-off between gain and speed—became a discipline in itself, and the small-signal models engineers built for it (the hybrid-pi model and its relatives) let circuit designers treat a transistor as a set of predictable parameters.

Germanium to silicon, and the planar process

Early transistors were germanium, which was easier to purify. Silicon had a wider band gap and could tolerate far higher temperatures, but it was harder to process. Throughout the 1950s, metallurgy and purification improved—zone refining produced silicon of extraordinary purity—and by the late 1950s silicon began to displace germanium, especially for military and aerospace use where temperature mattered.

The decisive manufacturing breakthrough was the planar process, developed at Fairchild Semiconductor in 1959 by Jean Hoerni18. Instead of building devices as mesas or grown junctions, Hoerni grew a protective silicon-dioxide layer over the wafer, cut windows in it photolithographically, and diffused dopants through the windows. The oxide both masked diffusion and passivated the surface, which had been the great source of unreliability18. Planar devices were stable, and—critically—the process lent itself to making many devices on one wafer and interconnecting them with evaporated metal8. Robert Noyce at Fairchild and, independently, Jack Kilby at Texas Instruments produced the first integrated circuits in 1958–59; Kilby's was a hand-built germanium assembly, Noyce's a planar silicon circuit suited to mass production. The integrated circuit made the "circuit" a manufactured object rather than an assembly of parts.

The MOSFET and the logic revolution

The field-effect transistor had been conceived long before it could be built910. Julius Lilienfeld patented a field-effect device in the 1920s and 1930s, but surface states—traps at the semiconductor surface—prevented it from working. The planar process, with its clean oxide interface, changed that. In 1959–60, Dawon Kahng and Martin Atalla at Bell Labs demonstrated a working metal-oxide-semiconductor field-effect transistor (MOSFET): a gate electrode separated from the channel by a thin oxide, controlling current by electric field rather than by injected charge. The MOSFET draws essentially no gate current, is simple to fabricate, and scales down beautifully.

For years the bipolar transistor remained faster and dominated high-performance circuits, while MOSFETs were used where low power and high input impedance mattered. The balance shifted as fabrication improved. Complementary MOS (CMOS)—pairing n-channel and p-channel transistors so that one is off whenever the other is on—drew almost no static power, and by the 1980s CMOS had become the standard for digital logic11. The microprocessor (Intel's 4004 in 1971, and the generations that followed) and memory chips were built in MOS. The bipolar transistor did not disappear; it remained the choice for high-speed analog and for power, and the two technologies coexisted, sometimes on the same chip6.

Scaling becomes a program

In 1965 Gordon Moore observed that the number of components per integrated circuit was doubling roughly every year (later revised to about every two years), and that this could continue for some time12. Moore's observation became a target and a self-fulfilling expectation: the industry organized itself around shrinking12. The physics of shrinking was worked out in the 1970s, most influentially by Robert Dennard and colleagues at IBM in 1974, who described "constant-field scaling": if every dimension of a MOSFET is reduced by a factor and the doping and voltages are scaled correspondingly, the device's electric fields stay constant and its behavior stays predictable while it gets faster and denser. This was the design rule that made each generation of chips a smaller, better version of the last.

Scaling drove enormous engineering effort in lithography, oxidation, diffusion, ion implantation, and etching—the process steps that turn a design into silicon. It also drove the study of reliability: hot-carrier degradation, oxide breakdown, and electromigration became design constraints. And it created a demand for power devices: rectifiers, thyristors, and later power MOSFETs and insulated-gate bipolar transistors (IGBTs), which switch large currents and voltages for motor drives, power supplies, and grid equipment. Power semiconductor engineering developed its own physics—breakdown voltage versus on-resistance, safe operating area, thermal management—because a device that must block kilovolts is designed very differently from one that switches a volt.

Compound semiconductors and light

Silicon is superb for electronics but a poor light emitter, because its band gap is indirect: electrons and holes recombine without efficiently producing photons. Compound semiconductors—alloys of elements from different columns of the periodic table, such as gallium arsenide (GaAs) and later gallium nitride (GaN)—have direct band gaps and can emit light efficiently. In 1962, working groups at General Electric, IBM, and RCA independently demonstrated visible and infrared light-emitting diodes from GaAs and its alloys. The laser diode followed, and by the 1970s GaAs devices were also the fastest transistors available, used in microwave and later in optical communication.

The optoelectronic thread ran through telecommunications: semiconductor lasers and photodetectors made fiber-optic links possible, and compound-semiconductor photodiodes and avalanche detectors became the receivers. In the 1990s, blue LEDs based on gallium nitride—work that earned Isamu Akasaki, Hiroshi Amano, and Shuji Nakamura the Nobel Prize in Physics in 2014—completed the primary colors and made white LED lighting practical. Compound semiconductors also became the basis of high-frequency and high-power electronics, and later of displays and solar cells.

The end of easy scaling and the nanoscale frontier

By the early 2000s, constant-field scaling was running out of room. Gate oxides became so thin that electrons tunneled through them; supply voltages could not fall much further; and power density threatened to melt chips. The industry responded with materials and geometry: high-k dielectrics to replace silicon dioxide as the gate insulator, metal gates, and then, around 2011, the FinFET—a transistor whose channel is a thin vertical fin, giving the gate control on three sides. Strain engineering, silicon-on-insulator, and ever more elaborate lithography (culminating in extreme-ultraviolet lithography) kept the curve going8.

Beyond these incremental moves lies a research frontier that is genuinely open. As transistors approach atomic dimensions, the questions change: how to control a channel a few atoms wide, how to reduce leakage, how to build memory and logic that do not rely on charge alone. Researchers explore tunnel field-effect transistors (which exploit quantum tunneling to switch with lower voltage), spintronic devices that use electron spin rather than charge, ferroelectric and phase-change memories, and devices built from two-dimensional materials such as graphene and transition-metal dichalcogenides. Some of these are demonstrations; some are in early production; none has yet displaced CMOS at scale. The field's history has been a long argument that the physics of doped semiconductors can be engineered into ever smaller, more capable devices—and the current research is a test of how far that argument extends.

Where the field stands

Semiconductor electronics today is organized around a few large enterprises and a deep bench of specialized ones: logic and memory fabrication at the nanometer scale, power devices for electrification, compound-semiconductor devices for light and high frequency, and a research community probing the limits of the transistor1. The tools of the trade—band theory, junction physics, doping, the planar process, scaling rules—are taught as a connected body of knowledge, and each new device generation is still designed by reasoning from them13. The open questions are the ones the history has been circling: how small a switch can be, how little energy it can use, and what materials will carry the next stage of the argument.

References

  1. Semiconductor device (introduction, Cat's-whisker detector, Diode, First transistor, Development of the transistor, Improvements in transistor design, Metal-oxide semiconductor, Applications), Wikipedia
  2. Semiconductor, Wikipedia
  3. Band gap, Wikipedia
  4. P–n junction, Wikipedia
  5. Depletion region, Wikipedia
  6. Bipolar junction transistor (History, introduction), Wikipedia
  7. Insulated-gate bipolar transistor (History), Wikipedia
  8. Planar process (introduction, Development), Wikipedia
  9. MOSFET, Wikipedia
  10. Transistor, Wikipedia
  11. CMOS, Wikipedia
  12. Moore's law, Wikipedia
  13. Electronic band structure, Wikipedia

The frameworks of Semiconductor Electronics

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1931-Present

    Solid-State Physics and Band Theory

    Why do some crystals conduct and others not?

    Fly to its tower in Nousfarer

  2. 1949-Present

    Doping and Junction Engineering

    How do impurities and junctions control current?

    Fly to its tower in Nousfarer

  3. 1950-1990

    Bipolar junction transistor

    How can a small current control a large one?

    Fly to its tower in Nousfarer

  4. 1959-Present

    Planar process

    How can many devices be made reliably on one wafer?

    Fly to its tower in Nousfarer

  5. 1960-Present

    MOSFET and CMOS Logic Design

    How can an electric field switch a transistor?

    Fly to its tower in Nousfarer

  6. 1962-Present
  7. 1965-2010

    Scaling Theory and Moore's Law Program

    Why do chips keep getting smaller and cheaper?

    Fly to its tower in Nousfarer

  8. 1970-Present

    Power semiconductor device

    How can a device switch large currents and voltages?

    Fly to its tower in Nousfarer

  9. 1974-2005

    Constant-field scaling

    What happens when every dimension of a transistor shrinks?

    Fly to its tower in Nousfarer

  10. 2000-Present

    Nanoscale and Post-CMOS Device Engineering

    What comes after the transistor stops shrinking?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Solid-State Physics and Band TheoryDoping and Junction EngineeringBipolar junction transistorPlanar processMOSFET and CMOS Logic DesignCompound Semiconductor and Optoelectr…Scaling Theory and Moore's Law ProgramPower semiconductor deviceConstant-field scalingNanoscale and Post-CMOS Device Engine…

Beyond the field

Where Semiconductor Electronics's ideas came from in other fields, and what they led to there: the strongest 12 of 70 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Semiconductor Electronics is a planet in Nousfarer

Its city has a tower for each of these 10 frameworks. Land, walk in, and climb one concept at a time.

Land on Semiconductor Electronics

Related fields

Fields its ideas are linked to

More in Electrical Engineering

See where Semiconductor Electronics sits on the Map of Human Knowledge