Page 2 of 3
Silicon to the rescue
In his previous job Kilby had been involved in a plan to place components on a ceramic base so perhaps this was the spark that gave rise to his good idea - to build everything on a single piece of silicon. To us with the benefit of hindsight it now seems less of a leap of the imagination but that is because we don't suffer from the preconception that a different material is needed for each type of component. Carbon makes resistors, ceramic and metal make capacitors, copper makes connecting wires and inductors and germanium or silicon make transistors. The idea of trying to make everything out of germanium, or the then new silicon, seemed a bit silly.
Kilby worked out that it could be done. The start was a silicon resistor - an undoped strip of silicon - then a silicon capacitor - a reverse biased diode. The beauty of the idea was the components could all be made on the same slice of semi-conductor material complete with transistors and wiring.
The first chip - an oscillator
When the TI holiday was over Kilby was given the go ahead to try the idea - but only after he demonstrated that each of the components he described could indeed be made. Working alone he hand carved a piece of silicon to make a resistor and used a standard PN junction to show that the capacitor worked. The next step was to make everything on a single chip. The material was Germanium and the internal connections between the components were made by hand-applied gold wires. This made the first working integrated circuit, a phase shift oscillator, looked like a bird's nest in miniature ... but it worked.
All on one chip
Noyce should have come to the single chip solution earlier than he did, earlier than Kilby. After all his firm, Fairchild, was making transistors in the modern way - a large wafer subjected to diffusion on both sides to make a PNP or NPN structure. The wafer was then chopped up, wires were connected to the slices and a can placed over each one. The idea of wiring up the transistors on the wafer should have occurred sooner but it took one other step forward.
Jean Hoerni, one of the other founders of Fairchild, came up with the idea of the planar transistor - a flat PNP junction insulated from the outside world by a layer of silicon dioxide. The planar transistor was invented to overcome problems of contamination but Noyce started to think about the way that metal strips could be laid down on top of the insulating layer to connect separate planar transistors together.
Noyce's chip design was sophisticated
After inventing the connection method he then went on to think about resistors and capacitors and discovered that it was possible to put it all together on one piece of silicon. He was six months behind Kilby but his version of the idea was arguably more complete. He sketched the design of an adder built in on a single chip - it would be a few years before the technology got that far. His concept of the integrated circuit is essentially one that would be recognised today and not at all like the slightly odd birds nest of wires of the Kilby version.