Texas Instruments 

Bell had now sold 35 licences world-wide to the new technology and suddenly every electronics company was keen to learn about the new transistor. Gordon Teal left Bell Labs and went to a small company based in Dallas. Texas Instruments was late into the field but with Teal they rapidly built a semiconductor lab and production facility. They were first in 1954 with the fabrication of a silicon transistor. All previous devices had been based on germanium and silicon, with a higher melting point was much better at handling power than germanium.

It is only chance that Silicon Valley didn’t grow in the wide open plains of Texas rather than California - but it didn’t. What they did do was demonstrate that any small company with enough enthusiasm could get years ahead of the established competition in the rapidly changing world of solid state electronics.

Shockley Semiconductor

At the same time that TI announced the first silicon transistors William Shockley left Bell Labs to start his own semiconductor business. At first he found funding difficult to obtain but eventually he secured the backing of Beckman Instruments and he set up Shockley Semicondutor in Palo Alto - the first silicon company in Silicon Valley. His reputation was so good that he attracted some of the best young talents available at the time including Gordon Moore, a physical chemist; Robert Noyce, a physicist and Jean Hoerni, also a physicist .

The prospects for the company should have been good but Shockley was not a good manager and he failed to focus on a product line. In particular he insisted on pursuing a four-layer pnpn avalanche diode which was way too far ahead of its time. In 1957 a group of Shockley’s engineers approached Beckman in an effort to oust Shockley as manager. Unfortunately Beckman couldn’t do very much because Shockley had just been awarded the Nobel Prize for his work on transistor action and was riding high on a wave of publicity.

The Traitorous Eight

The group of eight rebels met over a weekend and agreed to quit Shockley Semiconductor and offer their services as a team to the highest bidder. The group, which included Noyce, Moore and Hoerni,  was looking for employment but an investment banker heard about them and decided to make them a different offer. The banker knew that the Fairchild Camera and Instrument company was looking for a way of breaking into the semiconductor market and this seemed like a good start.

The group were offered backing to found the Fairchild Semiconductor Corporation - all they had to do was put up $500 each, the rest of the funding would be supplied. The parent company retained the right to buy the new firm for $3 million at any time in the next eight years.

The Traitorous Eight under the Fairchild logo

In 1957 the Fairchild Semiconductor Corporation opened for business just a mile or so down the road from Shockley - who referred to them as “the Traitorous Eight”. The eight had the chance to earn big money for the first time in their careers if only they could get it right. This video from the Computer History Museum has more of their story:

At first Fairchild just made the standard double diffusion transistors but the team were interested in anything that could be used to make money. At the time only Shockley and Fairchild were making semiconductors in Silicon Valley and the demand wasn’t that great.

Then came the shock of the Russian space success - Sputnik was launched in 1957 - and the American space effort swung into action and demanded new and better solid state devices. The loss of the traitorous eight more or less finished off Shockley’s company and it never recovered. It was bought by ITT and eventually closed in 1969.

Planar Process

Fairchild on the other hand made important breakthroughs. The first was the planar process. Because they had concentrated on the diffusion method of making transistors this seemed a natural, if difficult, step. The mesa (Spanish for ‘table’) process created a transistor by depositing flat layers of silicon like a cake, hence its name. It worked but it was very sensitive to surface contaminants. Jean Hoerni discovered that using a layer of doped silicon oxide protected the surface of the chip and made it possible to build flat devices by diffusing the impurities to make the base into the collector substrate. The flat structure and the silicon dioxide coating not only made the transistor more reliable, it also made adding metal connections much easier.  Fairchild announced the planar process in 1960 and it was very successful.

A planar transistor

At the time transistors had more or less taken over as computer components but nowhere was the absurdity of mass producing transistors on a single wafer of silicon and then cutting them up into individual units more ridiculous. Hundreds of thousands of transistors had to be wired together to build computers and this was labour intensive, error prone and unreliable.

A much better method would be to form the connections on the wafer and build complete working circuits on silicon.  This much was obvious but no-one really knew how to do it and many thought that it would be too expensive or difficult anyway.

The first solution was proposed by Jack Kilby working for TI but to be honest his solution was very much a miniaturisation of the existing manual wiring techniques. Although Kilby’s method lead to a long patent wrangle over the rights to the first integrated circuit, it was Fairchild’s methods that lead to the modern integrated circuit. Kilby actually wired up individual transistors on a wafer. Eventually he worked towards building a germanium based circuit using etching techniques that TI had acquired.

Meanwhile back at Fairchild, Noyce was slowly realising that their planar technique could be use to create other components than just transistors. One day he walked into Hoerni’s office and sketched how to create a resistor, the next day a capacitor, then how to isolate devices from one another using reversed biased diode junctions  and so on. Each day another piece of the jigsaw was slotted into place until Fairchild had the complete process.

The first RTL logic gate.

In 1959 Fairchild announced its line of RTL logic chips, the MicroLogic family. Soon after TI implemented Fairchild’s method of producing chips and launched its logic family, the 51 family  Although ridiculously expensive at first the integrated circuit logic gate was now the obvious way to build computers and Silicon Valley was in competition with Texas.

To see how all this led to the foundation of Intel and the range of chips, starting with the 4040, that dominated, the microprorcessor revolution, see Intel -  The Microprocessor Revolution.

<ASIN:0195338219>

<ASIN:0804737347>

<ASIN:0521827221>

<ASIN:0195163435>