The Invention of the Integrated Circuit : Jean Hoerni's Patent Notebook
This week marks 65 years since Jean Hoerni applied for a patent for his 'planar process' a key step towards the creation of the modern monolithic integrated circuit.
It’s often said that we stand on the shoulders of giants. Here we can look over the shoulders of two giants and see them working together.
Science writers are sometimes fond of ‘Eureka’ moments. The original legend is that Archimedes exclaimed ‘Eureka', Greek for ‘I have found it’ when he realized that the volume of water displaced from his bath must equal the volume of his own body. Archimedes reportedly used his discovery to test whether a golden crown had been debased with cheaper silver.
The authenticity of this story is dubious. A more authentic ‘Eureka’ moment is documented by mathematician Gauss writing EUREKA in one of his notebooks on the discovery of the proof of a theorem in number theory.
Gold mining in California led to Eureka becoming the state’s motto:
During the 1849 California State Constitutional Convention at Monterey, the delegates had to agree upon a design for the Great Seal of California, which included a representative motto. … Since the seal’s design included a gold miner, it seemed most fitting to the delegates to adopt this phrase as the State Motto. It became the official State Motto in 1963. Seventeen California locations bear the name Eureka.
Today’s post is about a Eureka moment in the 1950s that would be far more valuable for California, and for the world, than any gold find. It was a key step towards the development of the modern integrated circuit. Like Gauss’s discovery, it’s documented in a handwritten notebook.
Incidentally, I’m a big fan of Noted byJillian Hess, a Substack that shares the contents of notebooks written by a wide range of writers, artists, scientists, and much more.
The integrated circuit was invented twice. First, by Jack Kilby at Texas Instruments in September 1958, and then independently, a few months later, by Robert Noyce at Fairchild Semiconductor. Noyce’s monolithic version formed the basis for future development and commercialization of the integrated circuit.
A crucial earlier step in the creation of the monolithic integrated circuit was the development of the ‘planar process’ for manufacturing transistors by Jean Hoerni, a physicist colleague of Noyce’s at Fairchild. Hoerni had been born in Switzerland and educated there and in the U.K. After traveling to the U.S., he worked at Shockley Semiconductor before joining Noyce and Gordon Moore as one of the ‘traitorous eight’ engineers who left Shockley to found Fairchild Semiconductor.
Fairchild Semiconductor started operations on 1 October 1957 at unremarkable offices in Palo Alto.
Most of the ‘traitorous eight’ were grappling with getting the company started and with issues with making ‘mesa’ transistors, so-called because, like a geological mesa, they stood proud from the surrounding silicon chip.
Hoerni was diverted with something else though. Gordon Moore later wrote:
Where most of us were setting up the facilities and developing the early processes, Hoerni was drawing in his notebook.
At Fairchild, like other researchers, Hoerni kept ‘patent notebooks’ in which he methodically recorded his experiments and results, for use in subsequent patent applications. Hoerni was sketching his ideas for the ‘planar process’ which would create transistors embedded into the ‘plain’ of the chip.
Hoerni’s entries in the notebook where he documented the planar process start with a heading that provides a brief description of what is to follow:
December 1, 1957
Method of protecting exposed p-n junctions at the surface of transistors by oxide masking techniques
The next page has Hoerni’s illustration of a wafer with holes created where the oxide layer has been removed.
And below this, is the first cross-section illustration of Hoerni’s idea.1
Added oxide layer after the diffusion of the base impurity.
The added silicon dioxide layer would protect the exposed ‘p-n’ junction exposed at the surface of the transistor.
Next to the diagram is a short inscription made in a different hand:
Read and understood Dec 1, 1957
R. N. Noyce
Hoerni had shown his idea to Robert Noyce on the same day he drew it in his notebook.
It would be many months, though, before Hoerni and his colleagues could test the idea. According to Gordon Moore, they were delayed by the limitations of their photolithography:
Even at Fairchild, we did not attempt to make Hoerni’s planar structure for some time after he proposed it. The original photolithography we developed only made three masks in a set. To make Hoerni’s planar transistor took a set of four.
When they did manage to try Hoerni’s idea then, according to Moore, “the results were fantastic”. Particularly important was that this new approach led to transistors that were better suited to the demanding environments where one key customer wanted to use them: in intercontinental ballistic missiles.
We had an approach that would allow us to make the highly reliable transistors that were being specified for the Minuteman intercontinental missiles.
Fairchild would file Hoerni’s idea as this patent, No 3,025,589, dated May 1, 1959, with Hoerni as the sole inventor.
This was a big step forward. Hoerni published details of the planar process in an October 1960 paper Planar Silicon Diodes and Transistors where he highlighted one key advantage of his approach:
Another advantage of planar structures is that they are the natural first step in any integrated approach where several elements, transistors or diodes, are built simultaneously on the same piece of semiconductor.
An integrated approach!
But this wasn’t yet, quite, the monolithic integrated circuit. The illustration of the final state of Hoerni’s device in the patent shows three vertical wires attached to the surface of the ‘chip’. To create a working circuit these wires would have to be attached, probably by hand, in what would have been an expensive and time-consuming process and one that would probably prevent transistors from being made much smaller.
But Robert Noyce hadn't just read and understood Hoerni’s idea. He carried on thinking about it and soon contributed his own crucial idea: to add a layer of metal connecting the transistors as an additional stage in the manufacturing process. By doing so he invented the monolithic integrated circuit.
Noyce’s patent for a “semiconductor device-and-lead structure'“ No 2,981,877 was filed just three months after Hoerni’s.
The modern semiconductor industry is built on Hoerni’s and Noyce’s ideas. The Computer History Museum records what happened next and its importance:
Fairchild introduced the 2N1613 planar transistor commercially in April 1960 and licensed rights to the process across the industry. The billion-transistor integrated circuits of today rely on Hoerni's breakthrough idea. One historian has called it "the most important innovation in the history of the semiconductor industry."
By 1965 Fairchild was making ‘monolithic’ integrated circuits with 60 components and Gordon Moore felt confident enough to make his famous prediction of a doubling of the number of components every year.
I caught a shiver down my spine reading Hoerni’s original diagrams and Noyce’s comment and signature. It’s often said that we stand on the shoulders of giants. Here we can look over the shoulders of two giants and see them working together.
Robert Noyce died in 1990. Jean Hoerni in 1997. Jack Kilby was awarded the Nobel Prize in Physics in 2000 for the invention of the integrated circuit. Noyce, and I think Hoerni too, had they lived, would surely have shared that prize with Kilby.
Postscript
There is a label stuck on the inside cover of Hoerni’s notebook, instructing the researcher to:
Make all entries legibly, neatly and in ink. (Do not use pencil and do not use your notebook as a ‘scratch pad’.)
Even great technologists sometimes need reminding to keep their notebooks neat!
A big thank you to the Computer History Museum for making this notebook available. Several other Fairchild notebooks from this era are also accessible online.
Links to Hoerni’s and Noyce’s Patents
If you’ve enjoyed this then you might also enjoy this earlier post on Jack Kilby’s integrated circuit.
Happy 65th Birthday, Integrated Circuit
On 12 September 1958, Jack Kilby demonstrated the integrated circuit for the first time. It was an unassuming object. A seeming hodge-podge of germanium, aluminium, gold wires and glue. A simple oscillator circuit with a single transistor, a capacitor and three resistors.
The very first post on this Substack was a tribute to Robert Noyce and his later role in the creation of the microprocessor.
Robert Noyce And The Road To El Dorado
“Noyce knew exactly what he possessed in this integrated circuit, or microchip, as the press would call it. Noyce knew that he had discovered the road to El Dorado.” Tom Wolfe: The tinkerings of Robert Noyce “Don't be encumbered by history, jus…
And for more on some of the original sites where early silicon chips were created.
The First Intel 'Fab'
A recent edition of the excellent Asianometry newsletter has a fascinating article and video describing TSMC’s latest ‘Giga-Fab’. The article and the video are both highly recommended.
Hoerni was prompted to think about this approach after listening to a paper presented by Mohamed M. Atalla.
Shockley semiconductors were at the time trying to develop Germanium based transistors. The choice of silicon as the future of semiconductors by the folks at Fairchild was an important one.
Great read!
There should surely be some kind of memorial to Jean Hoerni in Geneva.