Cover of The Transistor, a book by Ketineni RamaraoFrom the book The Transistor
Chapters 15 and 24. Read the full book free as a PDF.

The idea: control current with a field

In a field-effect transistor (FET) the control terminal, the gate, does not inject current into the device. A voltage on the gate sets up an electric field, and that field opens or narrows a conducting channel between two other terminals, the source and the drain. Almost no current flows into the gate, so very little power is spent controlling the device. Because it uses a single type of carrier, it is also called unipolar. Bipolar transistors are covered in NPN versus PNP transistors.

What MOS means

The letters describe the physical sandwich: a metal gate, an oxide insulator and a semiconductor, usually silicon. In an enhancement-mode n-channel MOSFET, the source and drain are n-type regions in a p-type body. With zero gate-to-source voltage, the channel is normally off. A sufficiently positive gate-to-source voltage forms a conducting channel under the oxide. Under normal bias, electrons move from source to drain and conventional current flows from drain to source. Depletion-mode MOSFETs behave differently: they are normally on at zero gate-to-source voltage.

Cross-section of an enhancement-mode n-channel MOSFET: n-type source and drain regions in a p-type silicon body, with a metal gate separated from the silicon by a thin oxide layer.
An enhancement-mode n-channel MOSFET in cross-section. Original IOI diagram.

Why it took so long

Julius Lilienfeld filed patents in Canada and the US between 1925 and 1928 for a field-effect element that is now recognised as an early MOS-type device. It did not work in practice. Shockley's field-effect amplifier of April 1945 failed too. Bardeen's explanation was that electrons trapped in surface states shielded the semiconductor from the electric field. The attempt to understand that failure led, in December 1947, to the first transistor, which was not a FET at all.

The surface-state problem was solved at Bell Labs. In 1959 M. M. (John) Atalla and Dawon Kahng made the first successful insulated-gate FET by working with thermally grown silicon dioxide, in which they found that the surface states at the silicon-oxide interface were much reduced. The device was slow and addressed no pressing need of the telephone system, so Bell did not push it. A 1961 Kahng memo noted the "possibility of application in integrated circuits". Researchers at Fairchild and RCA saw the same advantage.

From curiosity to commercial

MOS devices reached the market in 1964. General Microelectronics and Fairchild offered p-channel devices for logic and switching, and RCA introduced an n-channel transistor for amplifying signals. N-channel devices are faster than p-channel but were harder to make. In 1963, Frank Wanlass and C. T. Sah showed that pairing p-channel and n-channel MOS transistors made logic that drew close to zero power on standby. Wanlass patented the idea, which is now called CMOS.

A round silicon wafer patterned with a grid of small rectangular chips, shining in rainbow colours against a pale background.
A 300 mm (12 inch) silicon wafer carrying a grid of finished chips. Every chip holds transistors, and the rainbow sheen is light reflecting from the patterned surface. Photo: Peellden, CC BY-SA 3.0, via Wikimedia Commons.

Why it won

MOS transistors are smaller and use less power than bipolar devices. The Computer History Museum reports that over 99 percent of microchips use MOS transistors, and in an April 2018 blog counted 13 sextillion transistors manufactured, 99.9 percent of them MOS. As designers scaled chips to hundreds of thousands of transistors, CMOS gave the best answer to the resulting power density, according to the museum.

Still changing shape

The gate has had to adapt as the transistors shrank. TSMC says its 2 nanometre technology, with its first-generation nanosheet transistor, began volume production in the fourth quarter of 2025. The gate, oxide and silicon sandwich is still the idea, even as the geometry changes. For the larger story, see how transistors work.

Quick answers

What is a FET transistor?

A field-effect transistor controls current through a channel using an electric field from a gate. In a MOSFET the gate is insulated from the channel; other FET types use a different gate structure. Very little steady gate current is needed.

What does MOSFET stand for?

Metal-oxide-semiconductor field-effect transistor, named for its sandwich of a metal gate, an oxide insulator and a semiconductor.

Who invented the MOSFET?

M. M. (John) Atalla and Dawon Kahng at Bell Labs built the first working insulated-gate FET in 1959, demonstrated in 1960. Julius Lilienfeld had patented the idea in the 1920s.

Go deeper in the book: The Transistor · PDF · Source Companion · Browse the category

Sources

Lilienfeld, Atalla and Kahng, the 1964 products, CMOS and the 99 percent figure are from the Computer History Museum pages. The surface-state explanation is from the CHM 1947 page. The gate-all-around remark follows Chapters 15, 24 and 39 of the book.

More from this book

History · From the book The Transistor

The history of the transistor

The transistor was not found by luck.

Explainer · From the book The Transistor

How transistors work

A transistor lets a small electrical signal control a much larger flow of current.

Explainer · From the book The Transistor

NPN vs PNP transistors: what is the difference?

An NPN transistor and a PNP transistor are mirror images.

Power · From the book The Transistor

What is an insulated gate bipolar transistor (IGBT)?

Logic transistors switch at very low voltages.