Junction Transistor
The junction transistor is the three-layer semiconductor amplifier that William Shockley predicted theoretically in 1948 and that Bell Labs first made work in 1950, built as a sandwich of alternating n-type and p-type regions, either n-p-n or p-n-p. It replaced the fragile point-contact transistor of 1947 because it could be grown, cut and manufactured by controlled processes, with behavior that followed from the bulk of the crystal instead of the accident of two metal needles. This entry concerns the invention and its three manufacturing routes, grown, alloyed and diffused; the physics of how the layers amplify belongs to the bipolar-junction-transistor entry.
Shockley's 1948 theory predicted a device that nobody had yet been able to build. Working out the behavior of junctions between n-type and p-type material, he saw that a thin middle layer between two junctions could let current injected at one junction be collected at the other, with a gain controlled by the width and doping of the layers. The theory appeared in detail in 1949. It was a rare case of theory running ahead of technology: the device required germanium far purer and more uniform than anything routinely available in 1948, and single crystals whose doping could be changed on demand.
The first working grown-junction transistor of 1950 depended on crystal growing. Gordon Teal, a chemist who believed that single crystals were necessary, and his colleague John Little built a crystal puller at Bell Labs and by 1948-50 were drawing single-crystal germanium from a melt. Morgan Sparks and Teal changed the dopant in the melt during growth, so that the crystal contained an n-p-n sequence, and Sparks made the first working device from a slice of such a crystal in 1950. Bell Labs announced junction transistors publicly in July 1951. Grown junctions were the proof of concept, with the layers laid down as the crystal grew.
Alloy and diffused junctions turned a laboratory demonstration into a product. In the early 1950s companies including General Electric and RCA developed the alloy-junction transistor, in which small pellets of a metal such as indium are melted into both faces of a thin n-type germanium wafer; where the metal dissolves into the crystal and regrows, it forms p-type regions. Alloy devices were cheap and made the first mass-market products, such as transistor hearing aids and radios, practical. Diffusion, developed at Bell Labs in the mid-1950s, instead drives dopant atoms into a surface at high temperature, which controls layer thickness far more precisely and gives higher-frequency devices. Diffusion is the ancestor of the planar process of 1959.
The junction transistor won because of manufacturability and performance, but it also left some myths behind. Compared with the point-contact device it had lower noise, higher and more predictable gain, higher power capability and long-term stability, and its characteristics could be tuned by design rather than by adjusting a contact. A common misconception is that a junction transistor is just two diodes wired back to back. It is not: the two junctions share a base region so thin that carriers injected from the emitter cross it before recombining, and two separate diodes would show no gain. Germanium, with its low band gap and leakage current that rises with temperature, soon gave way to silicon, which Teal's group at Texas Instruments made in 1954.
Its legacy is both direct and structural. Junction bipolar transistors still appear in power supplies, radio-frequency circuits, sensors and audio equipment, and they live inside certain integrated circuits that combine them with field-effect devices. The deeper inheritance is the process lesson: crystals grown from purified melts, junctions defined by controlled doping, and layers whose thickness is engineered. Modern chip fabrication, from silicon wafers to the doped regions beneath every processor transistor, is a descendant of those manufacturing steps.
Read the junction transistor as the moment when the transistor stopped being a fragile experiment with two metal points and became a layered crystal structure that could be manufactured by purifying, growing, alloying and doping.