Lithography Process Catalog¶
Four EUV lithography steps that turn a 300mm silicon wafer into hundreds of high-end chips. The Lithography room of Franklin walks the operator through the pipeline — the same pipeline TSMC, Samsung, and Intel run every leading-edge node on.
Every claim on this page is warranted by a substrate row. Atom IDs are listed alongside each step; the same content renders into the chat narration via FranklinMeaningRenderer in the operator's locale, and is the body of the corresponding franklin_ontology_facts row that the chat oracle reads on demand.
The 4 EUV steps in plain terms¶
Wafer Load — 300mm silicon onto the ASML stage¶
Substrate atom: narration_litho_wafer_load_teaching · FOF-180
What it does: A pristine 300mm silicon wafer (the size of a large dinner plate) is loaded onto the lithography scanner. This wafer will host hundreds of high-end chips, each potentially worth thousands of dollars when finished.
CALORIE: The starting state of every modern computer chip below 7 nanometers. The wafer itself costs ~$200; the value it carries when finished can exceed $50,000.
CURE: Heals the substrate-purity deficit. Defects at the wafer stage propagate through every subsequent layer. Without ultra-pure silicon and atom-clean handling, the next 50 process steps build defects on defects.
Resist Coat — 50 nanometers of chemically-amplified photoresist¶
Substrate atom: narration_litho_resist_coat_teaching · FOF-181
What it does: Spin-coats the wafer at 3000 rpm to deposit a uniform 50nm film of chemically-amplified photoresist. The resist is the film EUV light will pattern — wherever EUV strikes, the resist's chemistry changes.
CALORIE: Modern photoresists are designed to respond to 13.5nm EUV photons. The chemistry is what lets a wafer move from analog (uniform film) to digital (patterned circuits).
CURE: Heals the resolution-vs-sensitivity deficit. Older resists couldn't capture 7nm features; new resist chemistries (metal-oxide, EUV-MOR) close that gap. Without resist progress, EUV scanners would be wasted on resists that can't see what the light writes.
EUV Exposure — 13.5nm wavelength, ASML Twinscan NXE:3600D¶
Substrate atom: narration_litho_euv_expose_teaching · FOF-182
What it does: Extreme Ultraviolet light at 13.5 nanometer wavelength burns the chip pattern into the photoresist film. 50 kilowatt-class laser-plasma source. ~155 wafers per hour. Each scanner costs ~$200 million.
CALORIE: This is how silicon below 7nm is made. Apple Silicon, Nvidia H100/B200, AMD MI300 — every leading-edge chip exists because of EUV. TSMC, Samsung, Intel all run on ASML scanners.
CURE: Heals the Moore's-Law-stalling deficit. Without EUV, the entire compute industry stops scaling around 7-10nm. AI training, smartphones, electric vehicle controllers all hit a wall. EUV is the one technology that keeps the digital world growing.
Overlay Metrology — sub-nanometer alignment to previous layers¶
Substrate atom: narration_litho_metrology_teaching · FOF-183
What it does: After developing the exposed resist, scatterometry measures how precisely this layer aligns to the layers beneath it. A modern chip has 30+ lithographic layers; each must align to within fractions of a nanometer of the layer below.
CALORIE: The measurement that turns the lithography sequence into a working chip. Without overlay control, every subsequent layer prints in slightly wrong positions and the whole device fails.
CURE: Heals the yield-economics deficit. At the 5nm node, a 300mm wafer holds ~100 high-end chips. 1% yield loss costs ~$100,000 per wafer. Sub-nanometer overlay metrology is the difference between $billions in yield and $billions in scrap.
Why lithography matters as a Franklin domain¶
Lithography is the load-bearing technology of the digital economy. Every smartphone, every AI accelerator, every electric vehicle controller, every modern medical scanner exists because someone, somewhere, ran an ASML scanner on a 300mm silicon wafer in a fab clean room.
The Franklin Lithography room teaches the operator what each step actually is, what value each step produces, and what deficit each step heals if it works correctly. The cell's substrate-resident teaching is the same audit-defensible content the operator can ask the chat about: "what is ASML", "how does EUV work", "why does overlay matter" — all answered from franklin_ontology_facts FOF-180..183.
The wiki entries cross-reference the substrate atoms. The substrate atoms cross-reference the wiki. There is no drift between what Franklin says in chat, what appears in this wiki, and what the substrate records. One truth; multiple operator-facing projections.
Substrate references¶
| Layer | Where the truth lives |
|---|---|
| Per-step semantic predicate | meaning_atoms.atom_id for each narration_litho_*_teaching |
| Per-step en-US rendering | meaning_renderings.rendered_text keyed by (atom_id, locale='en-US') |
| Per-step operator-facing ontology fact | franklin_ontology_facts.body for FOF-180..183 |
| LITHO-001 game contract | language_game_contracts.contract_doc |
| Render audit | meaning_render_audit |
| Migration that seeded the teaching enrichment | cells/xcode/Sources/GaiaFTCLCore/NarratorSchemaV37.swift |
Federation-cosigned
This page's source is sealed in the GaiaFTCL federation manifest — page SHA-256 2007c51a9bb4bcc9…, manifest witness a090592e0609adc8…, signed 2026-06-02T18:58:22Z by cell gaiaftcl-mac-cell. Verify with gaiaftcl wiki sign --all and compare wiki-all-signatures.json.