The Integrated Circuit: How Kilby and Noyce Solved Different Parts of the Same Problem
A patent- and process-aware history of Jack Kilby's 1958 demonstration, Robert Noyce's planar interconnection concept, and integrated-circuit manufacturing.
In 1958, Jack Kilby at Texas Instruments demonstrated an oscillator whose active and passive elements were formed from one piece of semiconductor material and connected with fine wires. In 1959, Robert Noyce at Fairchild Semiconductor described an integrated circuit built around the planar process, using insulating oxide and deposited metal to interconnect devices on the surface. Kilby proved monolithic integration; Noyce supplied an architecture much closer to scalable manufacturing.
The “tyranny of numbers” was an assembly problem
As transistor circuits grew, engineers had to manufacture, test, place, and wire thousands of separate components. Every connection added cost, volume, parasitics, and a failure opportunity. Micromodules and hybrid circuits tried to standardize small components and packaging, but they did not eliminate individual assembly.
Kilby, newly employed at TI and working during a company shutdown, reasoned that resistors, capacitors, and transistors could all be fabricated from semiconductor material. On September 12, 1958, his hand-built germanium circuit oscillated. Its elements were on one sliver, while thin external wires made the interconnections. The artifact was persuasive precisely because it operated, even though its construction was not a high-volume recipe.
Kilby’s patent application, filed in February 1959, described miniaturized electronic circuits formed in a body of semiconductor material. TI pursued the concept commercially and in military systems, but wire-connected germanium devices were difficult to scale into dense, reliable arrays.
The planar process made the surface useful
At Fairchild, Jean Hoerni developed the planar process: silicon dioxide protected and passivated the silicon surface while photolithographic openings allowed controlled diffusion. Noyce recognized that the oxide could also insulate metal interconnections deposited across the surface. His 1959 patent application combined planar devices, p-n junction isolation, and surface metallization into a manufacturing scheme.
This contribution was not merely drawing wires differently. Photolithography could define many devices and their connections in repeated process steps across a wafer. Silicon dioxide worked naturally with silicon, and planar surfaces supported later layers and increasing complexity. Yield, contamination control, masks, diffusion, and packaging still had to be mastered, but the core topology was scalable.
Fairchild produced commercial planar integrated circuits, while TI and Fairchild contested patents and later cross-licensed technology. Patent chronology, working demonstrations, and manufacturing impact answer different “who invented it?” questions. Treating one inventor as having the complete modern IC in isolation erases Kilby’s proof, Hoerni’s process, Noyce’s interconnection insight, and the teams that made products.
Integration changed design economics
Early integrated circuits contained only a few components and cost more than discrete alternatives. Aerospace and military buyers valued low weight and reliability enough to fund learning. As yields rose, the ability to replicate many circuits per wafer drove costs down and complexity up.
Standard logic families let designers compose systems from packaged functions. Metal-oxide-semiconductor technology later enabled higher densities and lower power, and computer-aided design turned layouts too complex for manual drafting into manufacturable masks. The integrated circuit became both a device and a production system.
Kilby received half of the 2000 Nobel Prize in Physics for his part in the invention of the integrated circuit; Noyce had died in 1990 and Nobel Prizes are not awarded posthumously. The award does not settle all credit—it highlights why exact timelines and processes matter. The decisive transition was from assembling components to manufacturing functions and their interconnections together.
Related:
- Intel 4004: How a Calculator Project Produced a Commercial Microprocessor
- The Berkeley RAID Paper: Naming Redundant Disk Arrays Without Inventing Every Technique
Sources: