If there is a single component on which all of modern computing depends, it is the transistor. A current processor contains billions of them, each of which ultimately does the same job as an electrically controlled switch. The invention dates back to December 1947, in the research laboratories of the American telephone company AT&T, the Bell Telephone Laboratories.
The problem with vacuum tubes
Before the transistor, electronics was based on vacuum tubes. A tube is an evacuated glass envelope with a heated filament that emits electrons; a control grid regulates the flow of current, amplifying a signal or acting as a switch.
Tubes worked, but they had obvious drawbacks:
- they consumed a lot of power, above all to heat the filament;
- they produced large amounts of heat;
- they failed fairly often, like light bulbs;
- they were bulky and fragile.
ENIAC, with its thousands of tubes, clearly showed the limits of this technology. But the problem also affected the telephone network, which used tubes to amplify signals over long distances and mechanical relays in its exchanges. For AT&T, finding a smaller, more efficient and more reliable replacement was a strategic matter.
Semiconductors
Bell Labs had set up a solid-state physics research group led by William Shockley. The goal was to use semiconductors, materials such as germanium and silicon that conduct electricity better than an insulator but worse than a metal. Their conductivity can be changed by adding tiny amounts of other elements, a process called doping.
Shockley had hypothesised that an electric field applied to a semiconductor could control its conduction, but the first experiments failed. The theoretical physicist John Bardeen realised that the problem lay in the behaviour of electrons at the surface of the material.
The point-contact transistor
Working together, Bardeen and the experimental physicist Walter Brattain built a very simple device: a germanium crystal on which two very closely spaced gold contacts rested, made by cutting a thin strip of gold foil wrapped around a plastic wedge. In mid-December 1947 the device amplified a signal. On 23 December it was demonstrated to the laboratory's managers.
It was the point-contact transistor. The name "transistor", a blend of transfer and resistor, was chosen shortly afterwards. The public announcement came in June 1948.
The junction transistor
Shockley, who had not taken part directly in the decisive experiment, set to work on his own and in the following weeks developed the theory of a different and sturdier device, the bipolar junction transistor, made of layers of semiconductor doped in different ways. It was easier to manufacture reliably and became the basis of the transistors of the 1950s.
In 1956 Bardeen, Brattain and Shockley jointly received the Nobel Prize in Physics. Bardeen won a second one in 1972, for the theory of superconductivity: he is the only person to have received two Nobel Prizes in Physics.
From germanium to silicon
The first transistors were made of germanium, which however performs poorly at high temperatures. In 1954 Texas Instruments began producing silicon transistors, which were more stable. The same year one of the first transistor radios, the Regency TR-1, reached the market and made clear to the general public what this technology meant: a pocket-sized, battery-powered set instead of a bulky piece of furniture.
In computers, transistors gradually replaced tubes between the mid and late 1950s, giving rise to what is often called the second generation of computers: smaller, more reliable and cheaper to run.
The birth of Silicon Valley
In 1956 Shockley left Bell Labs and founded Shockley Semiconductor Laboratory in Mountain View, California, recruiting talented young researchers. His management style, however, proved difficult. In 1957 eight of those researchers, including Robert Noyce and Gordon Moore, left the company and founded Fairchild Semiconductor. Shockley called them traitors, and the group is remembered as the "traitorous eight".
From Fairchild, in the following years, dozens of semiconductor companies were born directly or indirectly, including Intel. That is how the area south of San Francisco became Silicon Valley.
How it works, briefly
For a computer scientist, a transistor can be thought of as a switch with no moving parts. In a field-effect transistor, the type used in modern chips, a voltage applied to a terminal called the gate allows or blocks the flow of current between two other terminals. By combining transistors appropriately you get logic gates; from gates you build adders and memories; from these you build a processor.
| Vacuum tube | Transistor |
|---|---|
| Heated filament | No heating |
| Evacuated glass tube | Solid crystal |
| Centimetres | Today a few nanometres |
| Frequent failures | Very long life |
In short
- The point-contact transistor was built by Bardeen and Brattain at Bell Labs in December 1947.
- Shockley developed the junction transistor soon afterwards; the three received the Nobel Prize in Physics in 1956.
- Transistors replaced vacuum tubes in computers between the 1950s and 1960s, making them smaller and more reliable.
- The departure of eight researchers from Shockley's company led to Fairchild, the first nucleus of Silicon Valley.
