Noyce Monolithic Planar Integrated Circuit
US 2,981,877Vapor-Deposited Aluminum Interconnects Over Thermally Grown Silicon Dioxide Passivation
How It Works: Step-by-Step Mechanical & Physical Breakdown
In 1958, computers were limited by the 'Tyranny of Numbers': circuits required millions of discrete transistors, diodes, and resistors hand-soldered together with tiny wires. If a single solder joint failed, the entire room-sized computer crashed. Jack Kilby at Texas Instruments created the first integrated circuit in 1958 by connecting components on a germanium bar with hand-glued gold flying wires. Robert Noyce at Fairchild Semiconductor made the definitive breakthrough that launched the computer age: he realized that by using the planar process with a glass insulating layer of silicon dioxide ($SiO_2$), aluminum wiring could be evaporated directly across the chip's surface, printing millions of transistors and their interconnections simultaneously in one solid crystal.
A single crystal silicon wafer undergoes thermal oxidation to form a tough, insulating layer of silicon dioxide glass (). Photolithography etches microscopic contact holes through the glass into active p-n junctions below. Aluminum metal is vapor-deposited across the entire surface and etched into flat microscopic wires that travel over the glass insulator, directly connecting transistors without a single loose wire.
Interactive Real-Time 3D Physics Simulation
Initializing 3D WebGL Physics Engine
Calibrating studio lighting, shaders & telemetry...
Detailed Component Architecture
1Thermally Grown Silicon Dioxide (SiO2) Passivation
A micro-thin layer of pure glass grown on the silicon surface by heating with oxygen.
Has high dielectric breakdown strength (), insulating aluminum wires from shorting against the underlying silicon p-n junctions.
2Vapor-Deposited Aluminum Metallization Leads
Thin flat aluminum stripes vacuum-evaporated over the oxide.
Forms low-resistance ohmic contacts () at exposed silicon contact windows while routing signals across the glass surface.
3Photolithographic Planar Etching
Using photoresist light masks and acid baths to define microscopic circuit geometries.
Enables batch-fabrication of thousands of identical microchips on a single silicon wafer simultaneously.
Governing Physical Equations & Principles
Why It Still Matters
A modern SoC is still Noyce's stack: oxide on silicon, contact windows, metal that is allowed to run over junctions. Damascene copper and low-κ dielectrics changed the materials, not the topology. Intel, which Noyce co-founded in 1968, still ships that topology by the billion.
Legal Claims Decoder (1 Numbered Claims)
The Historical Bottleneck
By 1958 a large computer was a reliability problem disguised as an electronics problem. A design with 10⁵ transistors implied on the order of 3×10⁵ soldered joints. Each joint was a failure site. The 'tyranny of numbers' (Jack Morton at Bell Labs popularized the phrase) said you could not wire your way to a million devices.
Why Prior Art Failed
- •Kilby's 1958 TI germanium bar still used gold flying wires for some connections.
- •Mesa transistors left junction edges exposed; contamination killed yield.
- •Hybrid modules (Minuteman, IBM SMS) only hid the wiring, they did not remove it.
- •Photolithography existed for transistors, not yet for chip-scale metal.
“Jean Hoerni's planar process (Fairchild, 1959) left a sheet of SiO₂ over the wafer. Noyce's January 1959 note asked the obvious next question: evaporate aluminum on that glass, etch it into traces, and open windows only where you want contacts. The oxide is both passivation and the printed-circuit board.”
Patent Wars & Legal Litigations
TI said Kilby's US 3,138,743 already covered a plurality of components in one semiconductor body. Fairchild said that claim did not teach planar surface metal running over oxide.
Interference and infringement dragged through the 1960s. In 1969 the Court of Customs and Patent Appeals credited Noyce with the planar interconnect. Kilby kept the body-of-semiconductor idea. Neither company could ship legally without the other.
TI and Fairchild cross-licensed. The industry treated both men as inventors of the IC. Kilby received the 2000 Nobel Prize in Physics; Noyce had died in 1990 and the Nobel is not given posthumously.
US 2,981,877 issued 25 April 1961. Planar TTL and then MOS memories made the flying-lead IC a museum piece within a decade. Noyce spent the 1980s at SEMATECH arguing that the same interconnect physics now needed a national process consortium.
Fairchild's 1961 micrologic parts, then Intel (Noyce, Moore, Grove, 1968), made the planar IC a product line instead of a lab trick. Moore's 1965 density essay is a yield and interconnect essay; it assumes this patent's wiring method.
Colleagues called Noyce the Mayor of Silicon Valley. The title stuck because he left Fairchild to start Intel and because he ran meetings without the East Coast suit hierarchy the Shockley refugees had walked out on in 1957.
- The 'traitorous eight' left Shockley Semiconductor in 1957 and founded Fairchild. Hoerni, Noyce, Moore, and Last were in that group. The planar IC is a Fairchild invention in the narrow sense and a Shockley-lab diaspora invention in the wide one.
- Early Fairchild metal was aluminum on SiO₂. Purple plague (Au–Al intermetallics) at the package bonds was a 1960s reliability crisis that this patent does not mention and that packaging groups spent a decade fixing.
- Gordon Moore's 1965 Electronics article counts components per dollar and per chip. The curve only exists if the wires are printed with the transistors.