Transistors explained – what they are and what they do
A transistor - a word combination of transfer and resistor - is a semiconductor device that amplifies or switches electrical signals, and it is one of the building blocks of modern electronics. Trillions of transistors are embedded in electronic devices on Earth and in space.
Why transistors are important
The transistor has been called the “workhorse of electronic technology“ and the “nerve cells of the Information Age”. Without them we'd still be living in the 1950s, electronically speaking: no portable computers, no game consoles, no electronic cameras, forget about smartphones, smartwatches and the internet, no GPS, no space telescopes or Mars rovers, no modern cars, no tiny hearing aids, and bulky black and white TV sets. Manufacturing, finance, health care, science and research, transportation – almost all aspects of modern life would be affected. Transistors are also used for high-frequency applications, such as the oscillator circuits used to generate radio signals.
Since their invention in 1947, transistors have become unimaginably small, as small as the width of a single strand of DNA! For instance, IBM’s latest 2-nanometer (nm) technology allows the company to cram a staggering 50 billion transistors onto a chip the size of a fingernail. Even commercially used devices like the current top of the line M2 processors used by Apple are chips built with 5-nm technology that already contain 20 billion transistors.
By far the most common application of transistors today is for computer memory chips and microprocessors.
What is a transistor?
Transistors are the active components of a computer chip (also called a microchip, an integrated circuit or IC), which can contain billions of these devices etched into their shiny surfaces. On the chip, transistors act as interconnected miniature electrical switches that can amplify electrical signals or turn a current on or off.
Just like any electrical switch, a transistor needs to do three things exceptionally well: allow the maximum amount of current to flow through when it’s on; allow little to no current to leak when it’s off; and switch on and off as quickly as possible to guarantee optimal performance.
This means that each transistor can be in two different states, storing two numbers – zero and one. With billions of transistors, a chip can contain billions of zeros and ones, sending, receiving and processing a remarkable amount of data.
History of the transistor
In the 1940s, vacuum tubes and electromechanical relays were widely used in the quickly growing telephone networks. Electromechanical relays had made fully automatic telephone dialing and switching a reality, but the relays had low speeds.
Vacuum tubes were widely used as diodes and triodes in the electronics industry of the time. The first computer that would use vacuum tubes, the Electronic Numerical Integrator and Computer, also known as ENIAC, was built in 1946. In fact, the ENIAC used more than 17 000 vacuum tubes for its operation, allowing signals to be sent and calculations to be performed more quickly through the use of electrical switching rather than the slower mechanical switching.
The problem with vacuum tubes was that they were not very dependable, and they were huge compared to transistors – as a consequence, the ENIAC occupied a 15x9 meter (50x30 foot) large room. It processed about 500 floating point operations per second (FLOPS). The theoretical performance of Apple’s M2 processor in its latest iPhones is rated at 3.6 Teraflops, i.e., 3.6 trillion FLOPS – and it fits on a fingernail.
In the 1940s at Bell Labs in Murray Hill, New Jersey, physicists John Bardeen, Walter Brattain, and William Shockley set about studying semiconductor materials to see whether they could create a durable alternative that might eventually replace vacuum tubes in telephone networks.
Bardeen, Brattain and Shockley had tested various combinations of p-type and n-type semiconductors under different conditions until they finally found a configuration that would allow a thin layer of semiconductor to regulate a large flow of current between two electrodes.
On December 16, 1947, they demonstrated the first working transistor, now known as the point-contact transistor (a feat for which they were awarded the 1956 Nobel Prize in Physics). The first transistor was roughly the size of a thumb.
Although the point-contact transistor was the first transistor invented, it never succeeded commercially because hard-to-control variations in the metal-to-semiconductor point contacts made it difficult to manufacture them reliably and with uniform operating characteristics.
Commercial transistors started to take off after in July 1951, Bell Labs announced the successful invention and development of the junction transistor.
By the end of the 1950s, bipolar junction transistors had almost completely replaced electron tubes in computer applications.
The transistor worked more efficiently than the vacuum tube and operated on less power. Not only did this revolutionize telephone networks and computer technology, but it also allowed computers to become smaller and more cost-effective and advance at a faster pace. Transistors became the steppingstone between vacuum tubes and modern computer technology.
The year 1971 saw the advent of the first microprocessor: Intel’s 4004, incorporating 2,300 transistors and the first memory. The Intel 4004 microprocessor circuit line width was 10 microns, or 10,000 nanometers. 40 years later, an Intel Core processor with a 32 nm processing die held 560 million transistors. Another 10 years later, by 2021, the top-of-the-line M2 processors used by Apple are chips built with 5-nm technology that contain 20 billion transistors.
The complexity of integrated circuits has grown exponentially, doubling every 2 to 3 years according to Moore's law, as transistors continue to become increasingly miniaturized.
Moore’s Law, postulated by Intel’s co-founder Gordon Moore in 1965, observes the empirical regularity that the number of transistors on integrated circuits doubles approximately every two years. This observation has held for more than 50 years now and drives the progress in computing such as the processing speed and the price of computers.
How do transistors work?
A transistor is a device for controlling, amplifying, and generating electrical signals in virtually all modern electronic devices. It relies on the electronic properties of a semiconductor material for its function to regulate or control current or voltage flow, amplifying and generating these electrical signals, and acting as a switch/gate for them.
The transistor works like an electronic switch that can turn a current ON and OFF. Its basic working principle therefore applies directly to processing binary code (0, the current is blocked, 1 it goes through) in logic circuits (inverters, gates, adders, and memory cells). But a transistor can also be turned partly on, which is useful for building amplifiers.
The role of semiconductors
Unlike conductors such as metals, which have numerous free electrons to carry an electric current, semiconductors like silicon and germanium have very few charge carriers. However, the addition of small amounts of certain impurities – a process called doping – can change the number of charge carriers. Doping modulates its electrical, optical and structural properties and as a consequence enables silicon to gain free electrons that carry electric current.
When a semiconductor has undergone doping, it is then referred to as an extrinsic semiconductor. In contrast, a semiconductor in a pure undoped form is an intrinsic semiconductor.
In silicon doping, there are two types of impurities according to which silicon semiconductors are classified: n-type, where electrons flow out of it, and p-type, where electrons flow into it. Either way, the semiconductor enables the transistor to function as a switch or amplifier.
n-type and p-type transistors
For example, when a tiny bit of phosphorus or arsenic is doped into silicon, a good semiconductor is obtained with the electrons donated by the phosphorus or arsenic acting as charge carriers. Semiconductors obtained in this manner are called n-type semiconductors, since the charge of carriers is negative.
A more remarkable type of semiconductor is formed when a small amount of boron or gallium is doped on silicon. Boron or gallium provides a positively charged carrier by stealing an electron from silicon. In place of the electron, a hole is left behind, and this hole can move about within the semiconductor, acting as a carrier of positive charge. These semiconductors are called p-type semiconductors.
Much of the technical importance of semiconductors stems from the interplay of holes and electrons. The material difference between n- and p-type doping is the direction in which the electrons flow through the deposited layers of the semiconductor. Both n- and p-type silicon are good (but not great) conductors of electricity. N- and p-type silicon are nothing amazing alone. However, when you put them together, interesting behavior is exhibited at the junction between the two.
The operation of junction transistors, as well as most other semiconductor devices, depends heavily on the behavior of electrons and holes at the interface between two dissimilar layers, known as a p-n junction.
What types of transistors are there?
Transistor design has evolved over time, from planar to FinFET to gate-all-around (GAA) transistors.
BJT and MOSFET
Typically, a transistor – which is based on the transport of electrons in a solid (i.e., a solid-state device) and not in a vacuum, as in the vacuum tubes of old – consists of at least three terminals for connection to an electronic circuit: the base, the emitter, and the collector, as they are called in the classical bipolar junction transistor (BJT). In modern switching applications that use field effect transistors (FETs), they are called the gate, the drain, and the source.
The source acts as the emitter filament of an electron tube; the drain acts as the collector plate; and the gate acts as controller. These elements work differently in the two main types of transistor used today: bipolar junction transistors, which came first, and the metal-oxide-semiconductor field-effect transistor (MOSFET).
When working as an amplifier, a transistor transforms a small input current into a bigger output current. As a switch, it can be in one of two distinct states – ON or OFF – to control the flow of electronic signals through an electrical circuit or electronic device.
Today, MOSFETs are the most common type of transistor. This is due to the unique features of this transistor, mainly low-power consumption and high manufacturing yield. It can be used as a switch and also to amplify signals in electronic devices.
This design is the classical bipolar junction transistor (so called because it has two p-n junctions) that brings key elements of a transistor on a two-dimensional plane, including the gate, which modulates conductivity through a channel, the source, through which drive current enters the channel, and the drain, where current leaves the channel.
The types of bipolar junction transistor are n-p-n and p-n-p transistors. The n-p-n transistor is made by placing a p-type material between two n-type materials. The p-n-p transistor is made by placing an n-type material between two p-type materials.
All these components are built on a base of the semiconducting material silicon. This transistor concept was industrialized in the 1950s and 60s and was highly suited for mass production and miniaturization.
Over time, engineers discovered that it’s possible to exert more control over the flow of current in the channel by raising the gate above the plane of the silicon, like a fin above water. Whereas the bipolar junction transistor is current controlled, the field-effect transistor (FET) is voltage controlled. Also, the FET is a unipolar device, which means that it is made using either p-type or n-type material as main substrate. Hence the current conduction of a FET is done by either electrons or holes.
The main advantage of a FET is that it has a very high input impedance, which is in the order of mega Ohms. Additional advantages are low power consumption, low heat dissipation - making FETs highly efficient devices. Consequently, the industry made the switch from 2D planar transistors to 3D fin field-effect transistors, abbreviated as FinFETs.
In FinFET transistors, the gate wraps around the channel on three sides of a silicon fin, as opposed to across its top as in planar transistors. This creates an inversion layer with a much larger surface area, which allows the gate to better control the flow of current through the transistor. This means that more current can flow through with less leakage, and a lower gate voltage is needed to operate the transistor. In addition, the vertical geometry of FinFETs made it possible for engineers to pack more transistors in a chip, advancing Moore’s Law even further. The result was a chip with better performance, lower power consumption and a leading status through the 2010s.
With the state-of-the-art chip nodes that the leading chip manufacturers are manufacturing, FinFETs are reaching the limit of how high fins can go and how many fins can be placed side by side to boost their current-carrying capacity without suffering from electrical challenges.
To further improve the control of the transistor channel, engineers found a way to replace the vertical fin with a stack of horizontal sheets, creating a new concept called gate-all-around field-effect transistors, which are shortened to GAA transistors, or GAAFETs.
Gate-all-around transistors use stacked nanosheets. These separate horizontal sheets are vertically stacked so that the gate surrounds the channel on all four sides, further reducing leakage and increasing drive current. This means superior electrical signals pass through and between the transistors, improving chip performance.
Additionally, chipmakers now have the flexibility to vary the width of the nanosheets to best suit a particular chip design. Specifically, wide nanosheets allow for higher and better drive current, while narrow nanosheets can optimize power consumption.
How transistors are made
The transistors on microchips are made by building up layers of interconnected patterns on a silicon wafer. This manufacturing process is a highly intricate undertaking involving hundreds of high-precision steps undertaken in ultra-clean rooms with complex and specialized machinery.
Since their early days, transistors were made with silicon (Si), which remains the most widely used semiconductor due to the exceptional quality of the interface created by silicon and silicon oxide (SiO2), which serves as an insulator.
Here is a summarized overview of the key steps:
The first step in creating a microchip is typically to deposit thin films of materials onto the silicon wafer. These materials can be conductors, isolators or semiconductors.
Lithography, or photolithography, is the critical step in the computer chip-making process. It involves coating the wafer with photosensitive material and exposing it with light inside a lithography machine.
3) Photoresist coating
To print a layer of a chip, the wafer is first coated with a light-sensitive layer called a ‘photoresist’, or ‘resist’, for short. It then enters the lithography machine.
Inside the lithography machine, light is projected onto the wafer through the reticle containing the blueprint of the pattern to be printed. The system’s optics shrink and focus the pattern onto the resist. Where light hits the resist, it causes chemical changes, recreating the pattern from the reticle in the resist.
5) Computational lithography
The reticle containing the pattern to be printed on the wafer sometimes needs to be optimized by intentionally deforming the pattern to compensate for physical and chemical effects that occur during lithography.
6) Baking and developing
After leaving the lithography machine, the wafer is baked and developed to make those changes permanent, and some of the photoresist is washed away to create a pattern of open spaces in the resist.
Materials such as gases are used to etch away material from the open spaces creating during the developing phase, leaving a 3D version of the pattern.
8) Metrology and inspection
Throughout the chip production process, the wafer is measured and inspected for errors. These measurements are fed back to the systems and are used to optimize and stabilize the equipment.
9) Ion implantation
The wafer may also be bombarded with positive or negative ions to tune the semiconductor properties of parts of the pattern before the remaining photoresist is removed.
10) Repeat, repeat, repeat
Steps 1-9, from deposition to resist removal, is repeated until the wafer is covered in patterns, completing one layer of the wafer’s chips. To make an entire chip, this process can be repeated up to 100 times, laying patterns on top of patterns to create an integrated circuit.
11) Processed wafer
In the final step in production, the wafer is diced into individual chips, which are encapsulated in protective packages.
There is an excellent map of the manufacturing process, from the initial growth of a silicon oxide film all the way to the final wiring layer generation processes here at the Toshiba website.
Who makes transistors?
Taiwan Semiconductor Manufacturing Co. (which produces all of Apple’s chips), Intel and Samsung Electronics are the only companies that can manufacture the most advanced microchips. The reason this club is so exclusive is the immense capital cost of building semiconductor fabrication plants (fabs or foundries) that produce these chips. For instance, TSMC is reportedly spending $34 billion on its latest foundry that will make future 2 nm chips.
However, these fabs couldn’t exist on their own. The advanced semiconductor industry is an intricate network of specialist companies from America, Europe and Asia.
Just to give you an idea of this ecosystem: Using sophisticated computer-aided design tools and software from Synopsys and Cadence, companies like AMD, Qualcomm, Intel, Apple and Nvidia excel at the design of the most advanced chips. Applied Materials develops and manufactures equipment used in various steps of the wafer fabrication process. ASML provides lithography. Zeiss SMT specializes in optical lenses, which draws the stencils on the silicon wafers from those designs, using both deep and extreme ultraviolet light. Lam Research, KLA and other firms provide various sophisticated and highly specialized wafer fabrication equipment.