The Last Codex
EN

Theme

Language

Transistor

A switch without moving parts

Date
1947
Era
Modern age
Rebuild stage
V Electric age
Difficulty

Plate

Legend
  1. 1Gold-foil contacts
  2. 2Plastic wedge
  3. 3Spring
  4. 4Germanium slab
  5. 5Metal base
  6. 6Leads of a junction transistor

Anatomy

Gold-foil points on the germanium: emitter on the left, collector on the rightGermanium slab on its brass base contact, and the plastic wedge held down by a leaf springLeads to the binding posts, from left to right emitter, base and collector; the dots move with the currentCentre-zero meters: the input signal (left) and the output across the load (right)With this device John Bardeen and Walter Brattain first obtained amplification at Bell Labs, in December 1947. A slab of n-type germanium is soldered to a brass plate, which serves as the base contact. A plastic triangle stands on its point on the germanium and a spring presses it down. Gold foil folded over the two lower edges was slit with a razor at the point, leaving two contacts 0.05 to 0.25 mm apart: the emitter on the left and the collector on the right. The emitter takes a small positive bias, 0.6 mA at 0.7 V; the collector is held near 40 V negative and passes 2 mA, so 1.4 mA comes in through the base. Holes injected at the emitter cross to the collector and change its current more than the emitter's own: with the collector voltage held steady at its bias, a change of emitter current gives 1.79 times that change at the collector. A 20 kΩ load lets the collector voltage swing, and the ratio falls to 0.87, but the signal on the collector is then 63 times the signal on the emitter and the power 55 times; from a 500 Ω source, Bell engineers measured 17 dB, which the model reproduces. Bell published these numbers for its Type A transistor of 1949, the first made in quantity, and none for the wedge device, so the figure uses them. They describe small signals, and the model treats the whole range of the control as small. The two meters show the signal alone, without the steady bias, as a scope behind blocking capacitors would: up to 50 mV on the left and 4 V on the right, for a 1,000 Hz tone slowed to one cycle every 2.4 s, and the dots on the leads move the way conventional current flows, faster where it is larger. The batteries and the load resistor are left out of the drawing.

A transistor is a small piece of semiconductor in which a weak current or voltage controls a much stronger current, so it can amplify a signal or act as a switch. John Bardeen and Walter Brattain made the first one at Bell Labs in December 1947, and chips now hold billions of them.

What it is

A transistor is a valve for electric current with no moving parts. A small current or voltage applied to one terminal controls a much larger current between the other two. Biased with a suitable steady current or voltage, it reproduces a weak signal with greater amplitude: that is amplification, the job it does in a radio or a hearing aid. Switched between fully conducting and fully blocked, it works as a switch, and a computer is built from billions of such switches.

The plate shows, on the left, the first transistor, the point-contact device Bardeen and Brattain built in 1947: a slab of germanium on a metal base, with a plastic wedge pressed onto it by a spring. A strip of gold foil runs over the tip of the wedge and is slit with a razor so that two gold contacts touch the germanium a fraction of a millimetre apart. On the right is a junction transistor of the 1950s in its small metal can, with three wire leads.

How it works

Transistors are made from semiconductors such as germanium or silicon. A pure crystal of these elements conducts poorly. Adding a tiny trace of another element, called doping, raises its conductivity sharply. Phosphorus or arsenic gives the crystal spare electrons, which makes n-type material. Boron or gallium leaves gaps, called holes, that behave like positive charges, which makes p-type material.

A triode vacuum tube with its glowing filament, grid and plate The vacuum tube did the same job before 1950, but it needed a hot filament, a glass envelope and far more power.

Where p-type and n-type material meet, a junction forms that lets current pass easily in one direction only. A junction transistor stacks three layers, n-p-n or p-n-p, named the emitter, base and collector. The base is very thin. In an n-p-n transistor, a small current into the base lets a much larger current flow from collector to emitter, often a hundred times larger or more. In a p-n-p transistor all the directions are reversed.

Most transistors made today are field-effect transistors (MOSFETs). There, a voltage on a gate, insulated from the silicon by a thin layer of oxide, creates or closes a narrow conducting channel underneath. Almost no current flows into the gate while it holds its state. In the CMOS circuits used in chips, a transistor draws power mainly when it switches, which is one reason billions of them can share a single chip.

How to build it

Modern transistors are made in factories that need ultra-pure silicon, clean rooms and machines worth billions. What follows is a historical overview of how the first ones were made, in the order a rebuilding workshop would follow: first a point-contact detector, then a point-contact transistor, and only later junctions. It is not a recipe; anyone who wants to see a transistor at work should use a manufactured one in a low-voltage circuit.

Get a suitable crystal. Point-contact devices need a piece of n-type germanium of good purity. In the 1940s it came from wartime work on radar detectors. Galena (lead sulphide), the “cat’s whisker” crystal of early radio sets, only makes a crude detector diode.

Prepare the surface. The surface is ground flat and then etched clean, because the contact behaviour depends on it. This step uses strong acids and belongs in a properly equipped laboratory.

Make the contacts. Two sharp metal points, or a gold foil split over a plastic wedge as in 1947, must touch the crystal very close together. A spring holds them under light, steady pressure. The base of the crystal is soldered to a metal block that forms the third terminal.

Form and test. Early makers “formed” the collector contact by passing brief pulses of current through it, which improved the device’s gain, then measured each transistor. Many failed, and only those that passed the tests were kept.

Move to junctions. Grown-junction transistors need crystals pulled from a melt with controlled doping, which requires a furnace with fine temperature control and an inert atmosphere. Alloy-junction transistors, made from the early 1950s, were produced instead by fusing small pellets of indium onto both faces of a thin germanium wafer. Both are industrial processes.

Safety. Etching acids burn skin and eyes; hydrofluoric acid, used to etch silicon, can poison the body through the skin with delayed symptoms, even from a small splash, and must be left to specialists. Germanium and silicon dust should not be breathed. Soldering irons cause burns, and old solder contains lead. Test circuits at low voltage; mains electricity can kill.

History and evolution

A 1950s pocket radio opened to show its transistors and battery The Regency TR-1 of 1954 needed only four transistors.

Engineers had wanted a solid replacement for the vacuum tube since the 1920s. Julius Lilienfeld filed a patent for a field-effect device in 1925, but no working prototype is documented. After the Second World War, Bell Labs set up a solid-state physics group under William Shockley to find an amplifier for the telephone network. Shockley’s field-effect design failed. John Bardeen traced the failure to electrons trapped at the surface of the crystal, and he and Walter Brattain then tried driving current through point contacts instead.

On 16 December 1947 their device amplified a signal. A week later they showed it to Bell Labs managers. The company announced the invention in New York on 30 June 1948, and the engineer John Pierce coined the name “transistor”. In Paris that same year, Herbert Mataré and Heinrich Welker built a similar device independently, which they called the “transistron”.

A silicon chip seen under magnification, with fine bond wires at its edges An integrated circuit puts many transistors and their connections on a single piece of silicon.

Point-contact transistors were fragile. In January 1948 Shockley conceived the junction transistor, which Bell Labs could produce from 1951. The Regency TR-1 pocket radio of 1954 used four germanium transistors from Texas Instruments and sold more than 100,000 units. Silicon, which copes better with heat, then replaced germanium.

In 1958 Jack Kilby at Texas Instruments built the first integrated circuit on a small bar of germanium. Mohamed Atalla and Dawon Kahng demonstrated the MOS transistor at Bell Labs in 1960. Bardeen, Brattain and Shockley shared the 1956 Nobel Prize in Physics. Since then transistors have kept shrinking, and the MOS transistor is now made in greater numbers than any other manufactured object.

Type a name, a date, a material, a domain…