← Mind Maps ·

Semiconductors & Microelectronics

Solid-state physics through the integrated circuit to modern fabs — the pioneers, the paradigms, the scaling limits

A mind map of semiconductors and microelectronics: the solid-state physics foundations; the transistor breakthrough; the integrated circuit and the planar process; microprocessors, memory, and logic; lithography from contact printing to EUV; the fabless-foundry revolution; and the scaling laws, physical limits, and geopolitics that define the modern industry. Named inventors, labs, processes, and products with dates across six branches.

Physics & the TransistorIntegrated Circuit & Planar ProcessMicroprocessors, Memory & LogicLithography & Process NodesFabless-Foundry ModelScaling Laws, Limits & GeopoliticsSolid-state physics foundationsMaterials and crystal growthThe 1947 breakthroughShockley Semiconductor and the Traitorous EightMOS and the field-effect transistorThe IC raceThe planar processFairchild's spinoff cascadeProcess innovationsOther early playersThe microprocessor eraThe RISC insurgencyDRAM, SRAM, and flashJapanese DRAM rise and fallAccelerators: GPU, TPU, NPUPhotolithography generationsEUV lithographyProcess nodes — historical and marketingAdvanced transistorsInterconnect and packagingTSMC and the foundry birthFabless design housesIDM survivors and pivotsThe equipment and materials chainMead-Conway and the design revolutionThe named lawsPhysical limitsEconomic structureTradeoffs that define the fieldGeopolitics and policyKarl Ferdinand Braun — crystal rectification, 1874Edwin Hall — Hall effect, Johns Hopkins 1879Julius Edgar Lilienfeld — first FET patent, 1925 (not built)Oskar Heil — field-effect patent, 1934Felix Bloch — electron wave states in crystals, 1928 (Bloch theorem)Alan Wilson — band theory of semiconductors, Cambridge 1931Boris Davydov — p-n junction theory, USSR 1938 (undercredited)Walter Schottky — metal-semiconductor barrier, 1938Russell Ohl — silicon p-n junction at Bell Labs, 1940Gordon Teal & Morgan Sparks — zone refining for pure germanium, Bell Labs 1950Teal — single-crystal silicon growth, Texas Instruments 1954Jack Scaff — crystal purification methods during WWIIBell Telephone Laboratories Solid State Physics Group formed, 1945William Shockley leads the group — Mervin Kelly sanctions itJohn Bardeen & Walter Brattain demonstrate point-contact transistor, Dec 23, 1947Germanium, two gold foil contacts, an OA amplifying triodeShockley — junction (bipolar) transistor theory, 1948Shockley — Electrons and Holes in Semiconductors textbook, 1950Nobel Prize in Physics, 1956 — Shockley, Bardeen, BrattainName "transistor" coined by John R. Pierce at Bell Labs, 1948Shockley Semiconductor Laboratory — Mountain View, CA 1955Shockley's management style drives eight engineers to quit, 1957The Traitorous Eight: Moore, Noyce, Hoerni, Last, Grinich, Roberts, Kleiner, BlankFairchild Semiconductor founded, Sep 1957 ($3,600 each in seed)Arthur Rock arranges Fairchild Camera & Instrument fundingMohamed Atalla & Dawon Kahng — MOSFET demonstration, Bell Labs 1959CMOS invented by Frank Wanlass at Fairchild, 1963Wanlass & Sah — "Nanowatt Logic Using Field-Effect MOS Triodes," 1963Herbert Kroemer — heterostructure transistor theory, 1957 (Nobel 2000, undercredited)Jack Kilby — germanium IC demo at Texas Instruments, Sep 12, 1958Kilby patent — Miniaturized Electronic Circuits, filed Feb 1959Robert Noyce — monolithic silicon IC concept, Fairchild, Jan 1959Noyce patent — Semiconductor Device-and-Lead Structure, Apr 1961Kilby-Noyce cross-licensing agreement, 1966Nobel Prize in Physics, 2000 — Kilby (Noyce died 1990)Jean Hoerni — planar process at Fairchild, 1959SiO2 passivation layer — the key enablerPlanar process makes Noyce's monolithic IC manufacturablePhotolithography with photoresist — Lathrop & Nall at Diamond Ordnance Fuze Lab, 1957Bob Schweikert — first working planar IC at Fairchild, 1960Jean Hoerni, Jay Last, Eugene Kleiner leave Fairchild, 1961 (Amelco)Charlie Sporck leaves Fairchild for National Semiconductor, 1967Jerry Sanders and others leave for AMD, 1969Moore & Noyce leave to found Intel, Jul 1968Don Hoefler — "Silicon Valley USA" Electronic News articles, 1971~65 companies tracked back to Fairchild alumni — the "Fairchildren"Beam-lead technology — Bell Labs 1960sSilicon-on-insulator (SOI) concept — 1960sIon implantation — early 1970s commercial adoptionSelf-aligned gate — Federico Faggin at Fairchild, 1968Silicon-gate technology enables the 4004 microprocessorTexas Instruments TTL 7400 series — 1964 (industry-standard logic)Motorola Semiconductor founded, 1955RCA CMOS 4000 series — 1968Mostek 4096 DRAM — 1973 (introduces address multiplexing)Intel 4004 — Ted Hoff, Federico Faggin, Stanley Mazor, Masatoshi Shima, Nov 19714-bit, 2,300 transistors, 10 μm process, 740 kHzIntel 8008 — 8-bit, 1972 (commissioned by Datapoint)Intel 8080 — 1974 (the first dominant microprocessor)Motorola 6800 — 1974; MOS 6502 — Chuck Peddle, 1975 ($25 price point)Zilog Z80 — Faggin + Shima, 1976Intel 8086 — 1978 (founds x86 architecture)John Cocke — IBM 801 RISC, 1974–1980David Patterson — Berkeley RISC I, 1981John Hennessy — Stanford MIPS, 1981ARM — Acorn RISC Machine, Cambridge 1985SPARC — Sun Microsystems, 1987RISC-V — open ISA from Berkeley, 2010Robert Dennard — 1T-1C DRAM cell, IBM 1966–1968Intel 1103 DRAM — 1 Kbit, 1970 (first commercial DRAM)SRAM cell — six-transistor, used for cachesFujio Masuoka — NOR flash at Toshiba, 1980; NAND flash, 19873D NAND — Samsung V-NAND, 2013; 200+ layers by 2024HBM (High Bandwidth Memory) — SK Hynix / AMD, 2013CXL (Compute Express Link) — 2019 consortium standardVLSI Project — MITI-funded Japanese consortium, 1976–1980Hitachi, NEC, Fujitsu, Toshiba, Mitsubishi as "Big Five"Japan captures 80% of DRAM market by 1986US-Japan Semiconductor Agreement, 1986 — quotas and floor pricesIntel exits DRAM, 1985 — pivots to microprocessorsSematech founded, 1987 — US industry-government consortiumSamsung becomes #1 DRAM producer, 1993 — Korea displaces JapanNVIDIA founded — Jensen Huang, Chris Malachowsky, Curtis Priem, 1993NVIDIA GeForce 256 — first "GPU" branding, 1999CUDA — parallel computing platform, 2006Google TPU v1 — 2015, announced 2016 (inference-focused)NVIDIA H100 — Hopper architecture, 80GB HBM3, 2022NVIDIA Blackwell B200 — 2024Cerebras wafer-scale engine — 2019 (single 46,225 mm² die)Groq LPU, Tenstorrent, Graphcore — specialized AI siliconContact printing — 1960s (photoresist + mask direct contact)Proximity printing — 1970sProjection printing — 1973 (Perkin-Elmer Micralign)Step-and-repeat (steppers) — GCA, 1978I-line (365nm) lithography — 1980s dominantKrF excimer laser (248nm) — 1990s dominantArF excimer laser (193nm) — 2000s dominantImmersion ArF (193i) — water between lens and wafer, 2004Double / quad patterning — 2010s workaroundsASML — Dutch spinoff from Philips, 1984EUV wavelength: 13.5nm (soft X-ray)EUV source — Sn droplet hit by CO2 laser, 50kWTSMC N7+ uses EUV, 2019 — first volume EUV nodeHigh-NA EUV (0.55 NA) — TWINSCAN EXE:5000, 2024ASML EXE:5000 costs ~$380M per toolOnly ASML makes EUV — load-bearing chokepoint10 μm (4004, 1971) → 3 μm (1980s) → 1 μm (1988)180nm (1999) — early CMOS scaling era90nm (2004) — when Dennard scaling effectively ends45nm (2007) — Intel introduces high-k metal gate22nm (2012) — Intel FinFET (Tri-Gate) first14nm / 10nm / 7nm — naming decouples from gate length5nm (TSMC N5, 2020) — first volume EUV3nm (TSMC N3, 2022) — FinFET limit; Samsung uses GAA2nm nanosheet GAA (TSMC N2, 2025; Samsung SF2)FinFET — Chenming Hu group at Berkeley, 1999Intel 22nm Tri-Gate — first commercial FinFET, 2011Gate-all-around (GAA) nanosheet — Samsung 3GAE, 2022Complementary FET (CFET) — post-2030 roadmapHigh-k metal gate — HfO2 dielectric replaces SiO2, 2007Copper interconnect — IBM introduces, 1997 (replaces aluminum)Low-k dielectrics — SiOF, SiOC to reduce capacitance2.5D packaging — TSMC CoWoS (Chip-on-Wafer-on-Substrate), 20113D packaging — through-silicon vias (TSV), HBM stacksIntel Foveros 3D stacking, 2019Chiplets / UCIe open interconnect standard, 2022AMD MI300A/X — first volume CPU+GPU chiplet APU, 2023Morris Chang — recruited by Taiwanese government from TI, 1985TSMC founded — 1987, with ITRI and Philips capitalPure-play foundry — no products of its own, only manufacturingTSMC IPOs in Taiwan (1993) and ADR (1997)TSMC market cap surpasses Intel, 2020TSMC Arizona fab — 4nm, announced 2020, first production 2025Cypress Semiconductor, LSI Logic, Cirrus Logic — early fabless, 1980sQualcomm — founded 1985; mobile modem dominanceBroadcom — founded 1991NVIDIA — fabless from founding, 1993AMD goes fabless — spins off GlobalFoundries, 2009Apple M-series — ARM-based, TSMC-fabricated, 2020Intel — IDM dominant until ~2015; struggles with 10nm 2015–2021Intel IDM 2.0 — Pat Gelsinger's return and foundry ambition, 2021Samsung — memory leader, logic foundry since 2005SK Hynix — DRAM and NAND; HBM leadership 2020sMicron Technology — founded 1978, DRAM + NANDTexas Instruments — pivots from logic to analog leadership, 1990sASML — EUV lithography monopolyApplied Materials — deposition, etch, CMPLam Research — etch and depositionKLA — metrology and inspectionTokyo Electron — tracks, etch, cleaningShin-Etsu, SUMCO — silicon wafer duopolyJSR, Tokyo Ohka, Shin-Etsu — photoresist oligopolyCarver Mead — Introduction to VLSI Systems textbook, 1980Lynn Conway — structured VLSI design methodology, Xerox PARC 1978MOSIS — multi-project wafer service, 1981 (academic access to fabs)Synopsys — founded 1986, logic synthesis leaderCadence — founded 1988, physical design toolsEDA industry: Synopsys, Cadence, Siemens EDA (Mentor)Moore's Law — Gordon Moore, Electronics magazine, Apr 1965Moore's 1975 revision — doubling every 2 yearsDennard scaling — Robert Dennard et al., IEEE JSSC 1974Koomey's Law — performance per joule doubles every ~1.57 years, 2010Wright's Law applied to cost per transistor — continues even as density slowsRock's Law — cost of a fab doubles every 4 yearsMakimoto's Wave — 10-year cycle between standard and custom siliconHuang's Law — GPU performance doubles every ~2 years (informal)Subthreshold slope floor — 60 mV/decade at 300K (Boltzmann limit)Dennard scaling breakdown ~2005 — power density plateausQuantum tunneling at gate oxides thinner than ~1.2nmWire RC delay dominates beyond 10nmSRAM cell scaling effectively stops ~3nmSingle-atom dopant variability at extreme nodesLandauer limit — kT ln 2 per irreversible bit erasureLeading-edge fab cost ~$20B (TSMC N2, 2025)Leading-edge mask set ~$50M at 3nmFewer than 3 companies at the leading edge by 2025: TSMC, Samsung, IntelMature-node foundries: SMIC, UMC, GlobalFoundries, TSMC matureSemiconductor industry revenue ~$600B, 2024Industry concentration — top 10 firms >60% of revenuePerformance vs. power — the post-Dennard multicore pivotDensity vs. yield — smaller transistors harder to make reliablyIntegration (IDM) vs. specialization (fabless/foundry)General-purpose CPU vs. domain-specific acceleratorMonolithic vs. chiplet — cost vs. yield tradeoffMemory bandwidth vs. compute — the "memory wall"Shoreline length vs. die area — packaging constraintUS Entity List additions — Huawei 2019, SMIC 2020US export controls on advanced chips — Oct 2022 BIS rulesRevised controls Oct 2023; further tightened 2024EU Chips Act — 2023 (€43B target)US CHIPS and Science Act — Aug 2022 ($52B)TSMC Arizona, Samsung Austin, Intel Ohio fabsASML DUV sales restrictions to China, 2023Japan's Rapidus — 2nm fab ambition, Hokkaido, with IBM partnershipTaiwan "Silicon Shield" doctrineSemiconductors &MicroelectronicsBrian Tighe · Mind Maps
Orbital mind map. Scroll to zoom, drag to pan, or use the buttons above (+ / − / 0 keys also work). Hover a node to highlight its path to the center and the subtree beneath it.

How to read this

The center holds the topic. The six branches fan out bilaterally — three on each side — each in its own color. Sub-branches nest three levels deep under each top-level branch. Hover a leaf to trace the path back to the center; hover a branch to see everything it contains.

This is the shape the topic has when you try to hold the whole field in your head at once. It is not an argument; it is a scaffold. The essays argue against or within scaffolds like this one.

More in the series