The Origin of Semiconductors: From Silicon Discovery to the Transistor Revolution

On December 16, 1947, in a corner of Bell Labs, John Bardeen and Walter Brattain carefully brought two probes to the surface of a germanium crystal. The gap between the probes was less than half the width of a human hair. In that moment, a small input signal was amplified and delivered as output. It was the moment the first point-contact transistor came to life.^[1]

Yet among those who first heard the news of the experiment’s success, their supervisor William Shockley was not included. Shockley learned of the breakthrough only later — and was furious. That sense of exclusion would go on to redirect the entire history of semiconductors in a wholly different direction.

The history of semiconductors is a story about materials, and at the same time a story about human ambition, jealousy, betrayal, and innovation.

A Strange Material: The Discovery of Semiconductors

In the early nineteenth century, scientists encountered deeply puzzling materials. They were neither conductors like copper or silver, through which electricity flows freely, nor insulators like glass, through which electricity does not flow at all — they occupied some ambiguous middle ground.

In 1833, the British chemist Michael Faraday discovered that the electrical resistance of silver sulfide actually decreased as temperature rose.^[2] In most metals, resistance increases with temperature. This material moved in the opposite direction. Faraday could not explain what it meant. The physics of the era had no framework for understanding the phenomenon.

In 1874, the German physicist Karl Ferdinand Braun used galena (lead sulfide) crystals to demonstrate that electric current could flow in only one direction — what is known as the rectification effect.^[3] This was the first systematic observation of diode behavior. Yet Braun, too, could not explain why it happened.

The term “semiconductor” began to be applied in earnest to these mysterious materials toward the end of the nineteenth century, but explaining their physical principles would require waiting for the emergence of quantum mechanics in the early twentieth century.

The Cat’s Whisker and the Age of Radio

In the early twentieth century, these strange materials found a surprisingly practical application: radio.

Early radio receivers required a rectifier to convert radio waves into audio signals. The most widely used method at the time was to touch the surface of a galena crystal with a thin metal wire. Because this wire probe resembled a cat’s whisker, the device was called a “cat’s whisker detector.”^[4]

The cat’s whisker detector worked — but it was temperamental. Finding the exact spot on the crystal surface where it performed well was itself a skill, and even a small vibration could shift the probe and cut off reception. In 1901, the Indian physicist Jagadish Chandra Bose obtained a patent for this device^[4], and this peculiar semiconductor component spread to millions of homes during the golden age of radio in the 1920s.

Yet no one could explain why the cat’s whisker detector worked. The physical principles of semiconductors would not be elucidated until much later — and that understanding would form the very foundation of the transistor’s invention.

The Age of Vacuum Tubes and Its Limits

Electronic technology in the 1930s and 40s was dominated by the vacuum tube. Vacuum tubes amplified signals, controlled current, and performed calculations. The ENIAC — the first general-purpose electronic computer, completed in 1946 — comprised 18,000 vacuum tubes and weighed approximately 30 tons.^[5]

The problems with vacuum tubes were clear. They were bulky, consumed enormous amounts of power, generated heat, and broke down frequently. ENIAC’s tubes burned out at an average rate of roughly one every few days during operation. On some occasions, technicians spent hours hunting through more than 17,000 tubes to find the one that had failed.^[5]

AT&T’s Bell Telephone Laboratories had strong motivation to solve this problem. Running a telephone network at national scale required thousands of signal-amplifying repeaters, and vacuum-tube repeaters carried astronomically high maintenance costs. In 1945, Bell Labs established a solid-state physics research group under Shockley’s leadership, with the explicit goal of developing a solid-state amplifier based on semiconductor physics.^[6]

Three People, One Invention — But Credit Was Not Equally Shared

William Shockley, John Bardeen, and Walter Brattain. These three men had entirely different temperaments and roles.

Shockley was a theorist and an ambitious man. He had already conceived the idea of a “field effect” — that applying an electric field to the surface of a semiconductor could control current flow inside it — as early as 1945. But experiment after experiment failed. Shockley could not find the answer to why his theory was not working in practice.^[6]

Bardeen was the one who understood why Shockley’s theory was failing. He grasped that a phenomenon known as “surface states” existed on the surface of semiconductors, preventing external electric fields from penetrating into the interior.^[7] He devised an experimental design that circumvented this barrier.

Brattain was an experimental physicist with exceptional manual dexterity. Translating theory into a working physical device was Brattain’s domain.

Between November and December 1947, Bardeen and Brattain conducted intensive experiments while Shockley was away. They attached two strips of gold foil to a germanium crystal at an incredibly close separation — one for the input signal and one for measuring the output. On December 16, 1947, the device worked for the first time. On December 23, an official demonstration was held before Bell Labs management. That date is recorded as the official birth of the transistor.^[1]

Shockley received the news late. The knowledge that he had not directly contributed to the central invention of the group he led cut him deeply. And that wound drove him toward a new invention.

Shockley’s Revenge: The Junction Transistor

The point-contact transistor was revolutionary, but it was not practical. It was extremely sensitive, difficult to mass produce, and prone to noise.

Shockley threw himself into solitary research over the Christmas holiday. In January 1948, he developed a theoretical framework for an entirely different structure from Bardeen and Brattain’s approach — a “junction transistor” (also known as a bipolar junction transistor), in which two types of semiconductor (n-type and p-type) were layered like a sandwich.^[8]

N-type semiconductors are materials in which electrons serve as the majority carriers; p-type semiconductors are materials in which holes — vacancies left by missing electrons — serve as the majority carriers. When these two types are joined in an n-p-n or p-n-p configuration, a small current applied to the thin layer between the two junctions (the base) can control a much larger current flowing from emitter to collector. This is the amplification principle of the transistor.^[8]

The junction transistor was far more stable than the point-contact transistor and far more suited to mass production. It was Shockley’s junction transistor that became the true foundation of the semiconductor industry. The 1956 Nobel Prize in Physics was awarded jointly to all three men^[9], but Shockley believed he had made the most important contribution. Bardeen and Brattain, meanwhile, were growing increasingly exhausted by Shockley’s domineering conduct.

The Revolt of Eight: The Origin of Silicon Valley

In 1955, Shockley left Bell Labs and established the Shockley Semiconductor Laboratory near his hometown of Palo Alto, California.^[10] He recruited the finest young physicists and engineers of the era. Gordon Moore, Robert Noyce, Jean Hoerni, Julius Blank, Victor Grinich, Eugene Kleiner, Jay Last, Sheldon Roberts — the men who would later be known as the “Traitorous Eight.”^[10]

Shockley was a brilliant scientist but a disastrous manager. He subjected his employees to constant suspicion, excluded them from important decisions, and withheld resources from any research direction that conflicted with his own intuitions. At one point he even compelled his staff to take lie detector tests.^[10]

In 1957, these eight men collectively left Shockley’s lab and, with funding from Fairchild Camera and Instrument Corporation, founded Fairchild Semiconductor. It was the first large-scale spinoff startup in the history of the semiconductor industry.

Shockley was incensed. He called the event a “betrayal.” Yet viewed through the lens of history, without that betrayal, the very concept of Silicon Valley might not exist as we know it. Fairchild Semiconductor became the parent company of dozens of subsequent semiconductor firms, and the companies spawned from them in turn gave rise to hundreds of startups.

Two Chips, One Invention: The Race for the Integrated Circuit

The transistor had succeeded in replacing the vacuum tube. But by the mid-1950s a new problem had emerged: the limitations of connecting transistors, resistors, capacitors, and other components with wires. Military missile guidance systems and computers required thousands or tens of thousands of components, and manually soldering them together was running up against the limits of cost and error tolerance.

The solution was to integrate all circuit elements onto a single semiconductor substrate. And two men, working independently, implemented this idea at almost exactly the same time.

Jack Kilby had only been at Texas Instruments for a few months in 1958 when he conceived the key idea. That summer, while all his colleagues had left for vacation, Kilby was alone in the lab — as a new employee, he had no vacation days to use. In that time, he realized the concept that resistors, capacitors, and all other circuit elements could be fabricated on a single germanium substrate.^[11] On September 12, 1958, the device worked for the first time.

Texas Instruments filed a patent in February 1959. That same year, Robert Noyce filed an independent patent at Fairchild Semiconductor. Noyce’s approach held a decisive technical advantage: using the planar process developed by Fairchild’s Jean Hoerni, it formed wiring directly on a silicon surface using a silicon dioxide (SiO₂) insulating layer.^[12] This method was far better suited to mass production and became the standard in semiconductor manufacturing.

The patent dispute dragged on for more than a decade. Courts ultimately recognized Kilby’s priority for the “idea” of the integrated circuit, and Noyce’s priority for the planar process — the key manufacturing method.^[11][12] Today both men are recognized as co-inventors of the integrated circuit. Kilby was awarded the Nobel Prize in Physics in 2000.^[11]

Moore’s Law: Prophecy or Self-Fulfilling Prophecy?

In 1965, Gordon Moore — then working at Fairchild Semiconductor — contributed a short paper to the trade journal Electronics.^[13] In it, Moore analyzed data from the period and identified a pattern: the number of transistors on an integrated circuit was doubling roughly every year. He predicted that this trend would continue for at least ten more years.

In 1975, Moore revised the prediction. He judged that the pace had somewhat slowed, with the doubling period extending to approximately two years.^[13] This is what we now call Moore’s Law.

What is remarkable is that this “law” is not a law of nature — it came to function more like a target that companies were expected to hit. Intel and other semiconductor firms built their development roadmaps around Moore’s Law and invested enormous research and development budgets to meet those targets. The prediction became a self-fulfilling prophecy that drove the industry forward.^[14]

Moore’s Law held with stunning accuracy for approximately fifty years. Intel’s first commercial microprocessor, the 4004, released in 1971, integrated 2,300 transistors. The most advanced chips of the 2020s contain tens of billions of transistors.^[14] The same area of silicon now holds tens of millions of times more transistors than it did at the beginning.

Yet physical limits are coming into view. As of the 2020s, transistor gate lengths have already shrunk to a few nanometers, a scale at which quantum tunneling can cause transistors to allow current to flow unintentionally. The semiconductor industry is attempting to work around these limits through three-dimensional stacking, new materials, and chiplet designs — but whether Moore’s Law can continue in its traditional form is a question on which even industry insiders disagree.^[15]

The Modern Semiconductor Industry and Geopolitics

Semiconductors today are not merely components. They have become a matter of national security and the central resource of geo-economics.

Cutting-edge semiconductor manufacturing is extraordinarily concentrated. The fabrication of leading-edge logic semiconductors at seven nanometers and below is effectively monopolized by Taiwan’s TSMC (Taiwan Semiconductor Manufacturing Company) and South Korea’s Samsung. The advanced chips produced by these two companies are core infrastructure for modern civilization, from smartphones and servers to automobiles and military equipment.^[16]

This concentration is a vulnerability. When the COVID-19 pandemic disrupted semiconductor supply chains in 2021, manufacturing industries worldwide — including the automotive sector — experienced severe production disruptions. A single automobile contains dozens to hundreds of semiconductor chips, and the absence of even one critical component brings the entire production line to a halt.^[16]

Through the CHIPS and Science Act of 2022, the United States appropriated approximately $52.7 billion in subsidies for the construction of domestic semiconductor manufacturing facilities.^[17] The European Union is pursuing similar policies through the European Chips Act of 2023. China has set domestic semiconductor independence as a national objective and continues to invest at scale, while the United States has imposed strict controls on the export of advanced semiconductor equipment and technology to China.

The germanium crystal experiment at Bell Labs in December 1947 now stands at the center of a geopolitical contest for technological supremacy between nations.

What Betrayal Produced

If Shockley had not felt excluded from Bardeen and Brattain’s invention, would he have developed the junction transistor so swiftly? If the Fairchild Eight had never left Shockley’s lab, would Silicon Valley look anything like it does today? If Noyce had not challenged Kilby’s patent, would the planar process have become the industry standard?

The history of semiconductors reads less like an account of remarkable individuals achieving breakthroughs through collaboration than like a record of how conflict, competition, and exclusion ignited innovation. Shockley’s autocratic behavior produced the departure of eight people; their departure produced the integrated circuit; the integrated circuit produced Moore’s Law. The personal experience of failure and rejection reshaped the structure of an entire industry.

When the probe moved across the germanium crystal in December 1947, it was not simply the moment a current was amplified. It was the beginning of a digital world — shaped over the following seventy-five years by human ambition and competition, cooperation and betrayal, all of it condensed and brought to form.