Join us for a deep dive into the tiny technology that powers the world.

Transistors are tiny electronic components that act as switches and amplifiers, and they dwell at the heart of modern technology. In simple terms, a transistor can turn a flow of electricity on or off, or boost a weak electrical signal into a stronger one. Built from materials with semiconductor properties (more on these below), transistors use a small input current or voltage to control a larger current between two other terminals. This ability to precisely control electrical signals is fundamental to nearly every electronic device we use today, from smartphones and laptops to data centers and solar inverters.

Modern computer chips contain billions of transistors etched onto a fingernail-sized piece of silicon. In fact, a typical smartphone processor now contains on the order of 15 billion transistors -- a number that would have sounded like science fiction to the engineers who invented the transistor in the 1940s.

A Brief History

The first transistor, invented at Bell Labs in 1947, was a handmade device about half an inch tall -- enormous by today's standards, when billions fit on a single chip. On December 23, physicists John Bardeen and Walter Brattain, working under William Shockley, demonstrated the point-contact transistor: a thumb-sized germanium crystal with two gold foil contacts pressed onto its surface to control electron flow. Though primitive in appearance, it marked the birth of solid-state electronics. (For this breakthrough, the trio would share the 1956 Nobel Prize in Physics.)

Following the point-contact transistor, in 1948 Shockley debuted the junction transistor, a more robust transistor made by "growing" (or fusing) semiconductor layers together. Transistors of the early 1950s primarily used germanium crystals. But by the latter half of the decade, silicon -- a more stable semiconductor, abundant in sand -- had begun to take over as the material of choice after the development of silicon purification and oxide layering techniques. Silicon transistors performed better at higher temperatures and soon supplanted germanium devices in most applications.

Why was the transistor such a big deal? Until the late 1940s, electronics relied on vacuum tubes, glass bulbs that used a delicate filament inside a vacuum to moderate the flow of electrons. Vacuum tubes were the functional units of the first radios and computers, but they were big, fragile, power-hungry, and prone to burning out. The transistor, in contrast, was tiny and made of a solid crystal material, with no filament needed.

The first electronic devices that could be called supercomputers used vacuum tubes, but they were enormous, sometimes taking up entire buildings. Transistors could switch on and off faster, last longer, and run on lower voltages than tubes. This meant electronic devices could be made much smaller, more energy-efficient, and more reliable, not to mention affordable. No longer were computers the province of universities or the military. Solid-state transistors, in turn, enabled technologies like the transistor radio, which took over a field once dominated by technologies like the cat's-whisker detector. Indeed, by the 1950s, transistors rapidly began replacing vacuum tubes in various consumer products such as hearing aids and calculators. The age of miniaturized electronics had begun.

A major breakthrough came in 1958, when Texas Instruments engineer Jack Kilby demonstrated the first integrated circuit (IC). Instead of wiring individual transistors together one by one, Kilby's approach was to build multiple devices on a single semiconductor slab, creating a complete circuit on a chip. In '59, Robert Noyce at Fairchild Semiconductor refined this with the planar process, making it practical to mass-produce chips with many transistors. These advances led to the first commercial microchips, which debuted in commercial products in the early 1960s.

By the late 1960s, integrated circuits were central to technologies like the Apollo 11 spacecraft's guidance computer, which used ICs instead of tubes. All those delicate, breakable bulbs and filaments were nonstarters at the level of "shake, rattle, and roll" vibrations they'd experience during launch and landing. The Apollo Guidance Computer used over 5,000 silicon ICs, making it one of the earliest large-scale transistor applications.

How Do Transistors Work?

At its core, a transistor is an electronic gatekeeper. It controls the flow of electric current through a semiconductor material. Most transistors have three terminals, often labeled as emitter, base, and collector (in bipolar junction transistors) or source, gate, and drain (in field-effect transistors). Applying a small signal to one terminal allows you to modulate a much larger current flowing through the other two.

To illustrate the idea, imagine a garden hose valve where the water flow represents electrical current. A slight twist of the valve handle (a change in the input signal) can completely start or stop the flow of water through the hose (output current), or anything in between. Now replace the mechanical valve with a semiconductor structure and the hand turning it with a voltage at the gate/base. That's essentially how a transistor modulates electrical flow.

There are two main families of transistors:

Bipolar Junction Transistors (BJTs): These were the earliest transistors (including the original 1947 point-contact device and later junction transistors). BJTs are made of three alternating semiconductor layers, either in an NPN or PNP arrangement, depending on the application. A small current into the base terminal allows a much larger current to flow from the emitter to the collector (or vice versa). BJTs can amplify analog signals or act as on/off switches.

Field-Effect Transistors (FETs): Instead of requiring a base current, FETs use an electric field at the gate terminal to control current flowing in a channel between the source and drain. The most critical subtype is the MOSFET (Metal-Oxide-Semiconductor FET), which was first demonstrated in 1960. A MOSFET has a gate electrode separated from the semiconductor channel by a thin insulating oxide. Applying voltage to the gate creates or destroys a conducting path in the underlying silicon channel. A positive gate voltage, for example, can attract electrons into a channel (for an n-channel MOSFET) connecting source to drain -- switching the transistor "on," and permitting the flow of electrical current. Remove the voltage, and the channel vanishes -- switching it "off."

Despite these technical differences, all transistors perform the same job: using a small input to control a larger output. In analog electronics, that means amplifying signals (e.g., boosting a microphone's electrical signal to drive a loudspeaker). In digital electronics, it means acting as fast switches, opening or closing to represent binary 1s and 0s. For example, a logic gate in a computer is typically built from CMOS transistors (paired MOSFETs) that output a high or low voltage depending on input conditions. Today's transistors can switch millions or even billions of times per second, enabling the astonishing speeds of modern processors.

Why Silicon?

Transistors are made from semiconductor materials, chiefly silicon. (A semiconductor is a substance whose conditional conductivity can be engineered and controlled.) Pure silicon is actually a lousy electrical conductor. However, by a process called doping, single-atom impurities are introduced to create two types of silicon with beneficial electrical properties:

* N-type silicon: Doped to have extra electrons (negatively charged).

* P-type silicon: Doped to have "holes" (locations that lack an electron, effectively positive charges).

Where N-type and P-type regions meet (a PN junction), an interesting electrical behavior emerges: Current can flow in one direction under certain bias, but not the other. It's a little like a subway turnstile, or Maxwell's daemon if you're familiar with the analogy. Transistors leverage multiple PN junctions. In a BJT, for example, an NPN transistor has a thin P-type base between an N-type emitter and N-type collector. A small current into the base-emitter junction allows a large current from the emitter to the collector. In a MOSFET, the source and drain might be N-type regions in a P-type substrate; a gate voltage inverts a thin substrate layer to N-type, bridging the source to drain with a temporary N-type channel.

The physics can get deep, but the key point is that by layering and shaping N and P regions, we create a controllable path for electrons. The transistor's design ensures that a tiny tweak (voltage or current) at the control terminal dramatically alters the conductivity between the other terminals. That's how transistors do both amplification and switching.

Silicon became the dominant semiconductor because it forms a tough oxide (silicon dioxide), an excellent insulator used as the MOSFET's gate dielectric. It is also abundant and cost-effective. Other materials are used too: Germanium was popular early on, and compound semiconductors like gallium arsenide (GaAs) are used for special applications (e.g., very high-frequency radio amplifiers or specialized microwave chips). In recent years, materials like gallium nitride (GaN) or silicon carbide (SiC) have emerged for high-power transistors (like those in electric vehicle inverters or RF power amplifiers) due to their ability to handle high voltages and frequencies. But for logic chips and CPUs, silicon MOSFETs have been the workhorse for decades.

Transistors In Integrated Circuits and CPUs

One transistor by itself is interesting, but the true power of the transistor shows itself when we put a ton of them together. This is where integrated circuits (ICs) and microchips enter the picture. The first transistor radio had four transistors. By the 1970s, engineers could put a few thousand transistors on one chip, enabling the first microprocessors (essentially tiny computers on a chip). Today's microprocessors pack tens of billions of transistors on a single piece of silicon.

In a CPU (central processing unit) or any logic chip, transistors are arranged to form logic gates (AND, OR, NOT, etc.), which in turn form complex circuits that perform arithmetic, storage, and decision-making. The prevalent design style is CMOS logic -- using complementary MOSFETs (pairs of N-channel and P-channel MOSFETs) to implement gates that consume very little power when idle. Each bit of data in a processor's registers or a memory cell is represented by the state of one or more transistors (on = 1, off = 0). By switching states billions of times per second, these transistor circuits execute software instructions, render graphics, and generally run our digital world.

Memory chips also use transistors: Static RAM (SRAM) uses a few transistors to store one bit in a flip-flop arrangement, while dynamic RAM (DRAM) uses one transistor plus a tiny capacitor to store a bit (with the transistor acting as a switch to charge or discharge the capacitor). In flash memory, transistors can trap charge (floating-gate MOSFETs) to remember data even when power is off. But in all these cases, it's the transistor that provides the means to store and manipulate binary information, by electrically isolating or connecting nodes within a circuit on command.

Modern Transistor Innovations: FinFETs, GAA, and Nanosheets

Packing more transistors onto chips and making them switch faster has required steady innovations in how transistors are built. For many years, improvements came from simply shrinking the dimensions of planar MOSFETs (the traditional flat transistors on the chip surface). However, as transistors shrank (gate lengths below about 30 nanometers), engineers ran into limits like current leakage. The transistor channels wouldn't fully turn off, and the power waste grew exponentially as devices shrank ever smaller, cramming more and more transistors onto the same unit area. To overcome this, designers had to change the transistor's architecture.

FinFETs (Fin Field-Effect Transistors) were the first significant departure from the OG planar design. Instead of having a flat channel under a gate, a FinFET stands the channel up as a 3D "fin" of silicon, like the dorsal fin of a fish, and wraps the gate around three sides of that fin. This geometry gives far better electrostatic control over the channel: the gate can pinch off current more effectively when turning the transistor off, reducing leakage and cleaning up the signal. The concept of multi-gate transistors had been explored in research for years, but from about 2011, it became a reality in mass production. For example, Intel introduced "Tri-Gate" transistors at the 22 nm node (~2012), becoming one of the first companies to use FinFETs in their CPUs. Soon, all leading chip manufacturers (TSMC, Samsung, etc.) adopted FinFETs for 14nm, 10nm, 7nm processes and beyond. FinFETs enabled continued scaling, offering faster switching and lower power than comparable planar transistors.

Still, the march of miniaturization continued. By the late 2010s, even FinFETs started approaching their limits as feature sizes neared the single-digit nanometers. The next step has been GAA, or Gate-All-Around field-effect transistors. In a GAAFET, the gate fully encircles the channel on all four sides, not just three. This is achieved by having the channel be a thin nanowire or nanosheet suspended in the gate material. If a FinFET is like a bridge with the gate on three sides, a GAAFET is like a tunnel that surrounds the channel. The idea is to maximize control over the channel, squeezing every bit of leakage out even at extremely short channel lengths.

There are different flavors of GAAFET under development, but one prominent approach is the nanosheet transistor. Here, instead of a single wire, the channel is formed by stacking flat, wide nanoscopic sheets (or ribbons) of silicon. Each sheet is wrapped by the gate, and multiple sheets can be stacked vertically to carry more current (thus boosting the drive strength of the transistor). This nanosheet design is sometimes also called a Multi-Bridge-Channel FET (MBCFET) by Samsung or RibbonFET by Intel, but these are all forms of GAA transistors using horizontally stacked channels.

The transition from FinFET to GAA is currently well underway, having kicked off in earnest during the COVID-19 pandemic. Samsung announced the first commercial 3nm GAA transistors in 2022, used in an ASIC chip for cryptocurrency mining. For their part, TSMC and Intel aren't far behind, planning GAA/nanosheet transistors in their upcoming 2 nm generation processes. IBM demonstrated an industry-first prototype 2nm chip in 2021 that used stacked nanosheet GAAFETs, managing to fit 50 billion transistors on a chip the size of a fingernail.

The driving reasons behind these new architectures are power efficiency and control. As transistors perform their work, you want them either fully on (for strong current) or fully off (no leakage). FinFETs approach the off-state by having more gate contact area than planar designs. GAAFETs approach the same problem by removing any "ungated" sides of the channel -- the gate touches 100% of the channel's perimeter. This design suppresses unwanted current leakage even at finescale levels, where quantum effects and other leakage mechanisms become problematic.

Beyond the constant evolution of transistor geometry, material science is always looking to improve transistors. Silicon may eventually be replaced or augmented by semiconductors offering higher speed in some transistors. Research is underway on using germanium or III-V compound semiconductors for channels in certain high-performance transistors, since they can have higher electron mobility than silicon. Other researchers are experimenting with 2D materials (like graphene or molybdenum disulfide) for future nano-transistors only a few atoms thick. Designs like these are still experimental, but they underscore that transistor innovation hasn't stopped -- we'd argue that it's branching out, as we approach the boundaries imposed by the laws of physics.

The Little Switch That Could

For a device so tiny (modern transistors are measured in tens of nanometers -- far smaller than a wavelength of visible light), the transistor looms large over our everyday life. Consider that every email you send, every GPS route you navigate, and every digital photo you snap all rely on countless transistor operations happening instantaneously behind the scenes. When you press the power button on your phone, billions of transistor switches flip to initiate the startup. It happens so reliably and quickly that we hardly think about it, and that's a testament to the transistor's robust engineering. Unlike the midcentury, with its finicky vacuum tubes, solid-state transistors can work steadily for decades without maintenance. They consume microscopic doses of energy for each switch, enabling battery-powered devices to run for hours on a charge.

Still, transistors have limits, and finding ways to improve them further is an active area of research. Moore's Law -- the historical trend of doubling transistor counts -- has slowed in recent years. Still, new packaging methods (like 3D stacking of chips and chiplets) and transistor innovations (like GAAFETs) are extending progress. The future of transistors may involve more specialization: different types of transistors optimized for logic, memory, power, or radio frequency tasks, all integrated in advanced systems. Future transistors might even leverage principles of quantum physics or operate with novel concepts, as seen in experimental quantum bits or spintronic transistors, although those are still far from mainstream use.

At the end of the day, a transistor is a simple concept -- a controllable switch -- that can be built in myriad ways. Its genius is in its scalability and versatility. One transistor on its own can amplify a signal; millions together can form a computer; billions can power artificial intelligence. From 1947's lab-bench contraption to the latest 2 nm silicon nanosheets, the transistor has continually reinvented itself while staying true to its essence. As we continue to demand more from our technology, the transistor will undoubtedly remain central, the unsung hero powering today's and tomorrow's electronics.