After the first transistors had been demonstrated, it was obvious that it was not a device that was easy to manufacture and use. Much transistor development was required.
For the transistor to be developed to make it a viable device, not only did the processes need to be developed, but so did the materials.
The transistor development is as much a part of the transistor story as any other.
Junction transistor development
The first point contact transistors were very unreliable and not suitable for manufacture. A more reliable and manufacturable idea needed to be developed.
Realising this problem Shockley himself developed the idea for replacing the point contacts with a proper p-n junction. This was not a just a passing thought because all the theory behind the operation of the new style junction transistor had been fully defined by calculations. Probably what is more surprising is that he developed the idea only a matter of weeks after his team had invented the first point contact device.
Even though Shockley was able to prove the feasibility of the junction transistor, it took somewhat longer for him to be able to make one in the laboratory. It was not until April 1949 that the first one was produced. He did this by dropping some molten p-type germanium onto some hot n-type. The resultant blob then had to be sawn down the middle to make the two p-n junctions. Using this very crude demonstration he was able to show that the device had both current and power gain.
During the 1950s very large changes occurred in the development of transistors. Much of this was brought about by the improvement in techniques for manufacturing and refining the raw materials.
In 1950 Teal managed to use the Czockralski process for manufacturing crystals of germanium. Later in 1952 Pfann demonstrated the process of zone refining. In this process an induction coil is passed along the crystal. As it moves down it melts the material causing the impurities to be transported to the end. In this way semiconductors with much lower levels of impurity could be produced.
The early stages of these developments enabled Shockley to produce crystals which he could dope with controlled levels of the correct impurities to make a p-n junction. In the next stage of his work he managed to produce an complete n-p-n structure on a germanium crystal. Although the transistor worked, its performance was not up to the standard Shockley had anticipated. Further material development was required.
As the technology for the materials developed more companies started to produce transistors. Initially Bell marketed both point contact and "grown junction" varieties. Shortly afterwards General Electric introduced a type they called their alloyed junction device.
Up until this time all the transistors had been made from germanium. In fact at the conference of the Institute of Radio Engineers in May 1954 a number of speakers stated that silicon transistors were many years away from becoming a reality. To their surprise Teal who had now moved to a little known company called Texas Instruments, produced a working silicon device. This gave Texas a lead in the transistor field making them a major semiconductor manufacturer. It took the other manufacturers several years to introduce their own varieties of silicon device.
Whilst Texas took the lead by developing the silicon transistor, Bell and General Electric continued with other lines of investigation. The process of diffusing impurities into the semiconductor was perfected so that the required structures could be produced. In addition to this another major improvement was the ability to be able to grow oxide layers onto the structures. With photographic techniques this enabled the areas of impurity to be controlled more accurately.
With diffusion and photographic techniques established it was possible to produce many transistors from on single wafer. The slice could then be cut to produce the individual transistors. In this way they could be produced in sufficient quantities to start to bring the prices down to levels where they were more competitive with valves.
Despite the improvements, transistors were still comparatively expensive. By the early 1960s an ordinary marked transistor would cost around £1 10s 0d (£1-50), but for those not quite so bothered about specification, red and white spot transistors could be bought. These were effectively manufacturers rejects, but ideal in many ways for the amateur enthusiast. The colour denoted the frequency response of the device, red spot were audio devices and white spot ones were for r.f. applications, but with a frequency response limited to two or three MegaHertz at best. These devices could be bought much more cheaply at around five shillings (25p) each.
During the 1960s the prices of transistors fell dramatically as their use increased. Silicon replaced germanium and also the performance improved with transistors like the BC107 being in widespread use.