AI summary by TwinMind
Japan’s Legendary Semiconductor Breakthrough
Asianometry
21m 24s
Background and IBM Threat (1970-1975)
- Japanese computer market share declining: 60% (1970) → 48% (1974)
- IBM gaining ground with 30% market share, System 370 popularity
- IBM revenue: 4 trillion yen, R&D: 300 billion yen globally
- IBM held 62% world computer market vs Japan's largest (Fujitsu): 240 billion yen revenue, 25 billion yen R&D, 3% global share
- US government pressure on Japan to liberalize computer market
- Fear of complete IBM domination among Japanese politicians and MITI
The IBM Future Systems Rumor (1975)
- Fujitsu employee received IBM court documents revealing "Future Systems" (FS) plan
- IBM's secret 1971 plan to replace System 370 with revolutionary systems
- Target: 1 megabit memory chip by 1980
- IBM had already killed FS project due to program incompatibility issues
- Some ideas became System 38
- Rumor spurred Japanese government action on fourth-generation VLSI systems
Computer Generation Classifications
- First generation: vacuum tube
- Second generation: transistor
- Third generation: integrated circuit
- Generation 3.5 (System 360): LSI (large scale integrated)
- Fourth generation: VLSI (very large scale integrated) - over 100,000 devices
- Starting with 64k DRAM generation
Project Structure and Origins
- Initial idea: merge 5-6 Japanese computer companies (rejected by companies)
- Previous forced partnerships in late 1960s failed
- April 1974: Japan Electronic Industry Development Association proposed cooperative R&D
- Used Engineering Research Association (ERA) structure, adapted from UK research associations
- ERA model used in Japan since 1955, starting with radiator companies
- Applied to diverse industries: camera lenses, pistons, bearings, polymers, camshaft bearings
ERA Structure Features
- MITI issued formal tender to 10-12 companies
- One company designated as project lead
- Formal legal entity established
- Contract with MITI
- Dissolves after completion
VLSI Project Setup (1976)
- Timeline driven by IBM rumor: Japanese needed system ready by 1979 (vs IBM's 1980 target)
- Five participants: Fujitsu, Hitachi, Mitsubishi Electric, NEC, Toshiba
- Fierce competitors continuing battles outside project
- Focus on "basic and common" upstream problems
- Two main themes: micro fabrication and silicon wafers
- Goal: VLSI-class computers, not necessarily chips
- Excluded companies like Oki, Sony (not computer makers)
- Public announcement: March 1976
Budget and Management
- 4-year duration, 28 billion yen total budget
- 50% consortium participants, 50% government (interest-free repayable loans)
- Board of directors: presidents of five member companies
- Managing director: Masato Nabashi (MITI executive, later worked at IBM Japan)
- 80% funding to individual company laboratories
- 20% for cooperative efforts including joint lab
Laboratory Structure
- Full joint cooperative lab: NEC building wing, staff from all five companies
- Director: Yasu Tarui (94 years old as of writing)
- Two group labs: NEC/Toshiba partnership, other three companies partnership
US Tensions and Secrecy Issues
- May 1977: Variable shaped beam lithography device results published in Japanese newspaper
- US semiconductor leaders (Wolf Corgan/Fairchild, Charles Spork/National Semiconductor) called project anti-competitive
- December 5, 1977: Tarui asked to speak in Washington DC
- Tarui requested permission to clarify project scope and budget (10-20% for cooperative work)
- Managing director Nabashi denied permission
- Missed opportunity for US-Japan relations clarification
Joint Lab Technical Work - Lithography
- Six prototype tools produced: UV light, X-rays, electron beams
- Photolithography: optical systems using light to project designs onto wafers
- Challenge: 1 megabit memories required sub-micrometer features
- Japanese belief: photolithography couldn't achieve sub-micrometer due to wavelength limits
- Focus on X-ray and electron beam lithography as photolithography successors
X-ray Lithography
- Proximity x-ray lithography machine (shadow printing)
- High energy x-rays, no size reduction
- Mask same size as final product
- Found impractical due to mask-wafer alignment difficulties
Electron Beam Lithography
- Most promising next-generation technology
- Two directions explored:
- Electron beam projection lithography: beam onto mask, shrinks features to 1/4 size, 0.2 micrometer resolution
- Photo-cathode: light onto cesium iodide layer emits electrons, focused via magnetic field
- Electron beam direct lithography: beam manipulated like pencil to draw directly on substrate
- Devices transferred to Toshiba Machine (later New Flare Technology) for photo mask writing
Photolithography Development
- Despite focus elsewhere, photolithography work continued
- Established Nikon and Canon dominance
- Three exposure tools commissioned:
- VSR1 (Canon): 436nm G-line UV, proximity aligner, 1:1 mask, 2 micrometer features, 30x30mm plot
- VSR2 (Nikon): 10x reduction stepper, 1 micrometer features, 10x10mm plot, laser alignment system
- MR1: Deep UV (DUV) machine using mirrors, later sold by Canon for LCD panels
Stepper Technology Breakthrough
- VSR2 step-and-repeat tool required micrometer accuracy positioning
- Implemented laser position alignment system (patented by Electric Research Institute 10 years earlier)
- Unprecedented overlay error reduction
- Nikon commercialized VSR2 as NSR-1010G stepper (1980)
- Canon improved PLA series proximity aligners, entered stepper market with FPA-1500 (1984)
Silicon Crystal Technology
- 5-inch wafer prototypes from Japanese manufacturers
- Inspection for warpage and defects, working with manufacturers for improvements
- Epitaxial growth using beams for pure, well-structured silicon layers
- World-class silicon wafer work
- Shin-Etsu (non-VLSI member) became world's biggest silicon wafer maker
Other Technical Areas
- Dry etch: engraving chip designs without chemical wet baths
- Diagnosing: mask inspection tools, wafer production inspection
- Design structures: basic transistor design, CAD software, simulation technologies
Project Outcomes and Impact
- No single commercialized product produced
- Upgraded Japan's semiconductor equipment industry
- Strengthened NEC and Mitsubishi (weaker computer makers)
- 1980: Japanese companies invested $1 billion in new 64k memory plants
- 1982: Japanese achieved 70% market share in 64k memory (first VLSI systems)
- 1983: Unveiled 256k DRAM, high volume production
- 1985: Fujitsu achieved 1 megabit memory chip
US Response and Emulation
- 1983 US House Representatives report: coordinated Japanese research gave major 64k RAM market share
- US government/industry studied and emulated VLSI project
- National Cooperative Research Act of 1984 amended antitrust laws
- Enabled US semiconductor R&D consortiums
Project Assessment
- 80-90% funding spent conventionally in individual labs
- First semiconductor R&D project with cooperative portion
- Breakthrough copied by US and others
- Cooperation challenges: learnings didn't transfer between companies
- NEC Executive VP Muki Wanoara: "cooperation among competing companies almost impossible as technical development moves closer to product levels"
- Excellent timing, especially in lithography technology
- Major lasting effects on Japanese and global microelectronics industry
