Is the EUV mirror advantage genuinely tacit institutional know-how (not IP), and is the sub-0.1nm RMS roughness figure accurate?

This capture answers question-verify-euv-mirror-tacit-knowledge-moat, which flagged claim-euv-mirror-advantage-is-tacit-know-how-not-patent's two load-bearing claims — the causal mechanism (Nikon/Canon couldn't enter EUV because the ASML/Zeiss advantage is tacit institutional know-how, not patents) and the quantitative figure (mirrors polished below 0.1nm RMS roughness) — as resting entirely on a single Tier-4 industry explainer, below the sourcing floor for both claim types.

Bottom line: the quantitative figure checks out against a primary source; the tacit-knowledge framing is real and Tier-2-sourced as a general thesis about ASML/Zeiss's moat, but the specific causal story that Nikon and Canon tried and failed to replicate that know-how is not what the better-sourced accounts describe. The mechanism, as applied specifically to Nikon and Canon, should be treated as a documented exclusion-then-exit story, not a documented technical failure to imitate tacit know-how.

Claim: Carl Zeiss SMT's own 2011 technical presentation confirms a mirror surface-figure specification of rms = 0.1nm, corroborating the order of magnitude in the flagged quantitative claim

Claim type: quantitative. Source tier required: Tier 1-2. Source tier met: Tier 1.

A slide deck presented by Carl Zeiss SMT engineers (Olaf Conradi, Peter Kuerz, and co-authors) at the 2011 International EUV Lithography Symposium states, for the NXE:3300 projection optics then in production, "achieved surface quality rms = 0.1 nm" against an internal target for "surface figure." The same deck's mid-spatial-frequency roughness (MSFR) chart shows measured values in the 52–102 picometer RMS range across mirror generations from the NXE:3100 (2004–2009) to the first NXE:3300 mirrors (2011), with the best ("Champion") 3100 mirror at 59 picometers and the first 3300 mirrors at 52–78 picometers — all comfortably at or below 0.1nm (100 picometers) RMS.

This is a primary, named, company-authored technical document — the people who built the mirrors, reporting their own fabrication data — and it clears the Tier 1-2 floor required for a quantitative claim. It corroborates the order of magnitude of the "sub-0.1nm RMS roughness" figure originally sourced only to semiconductorx.com. Two caveats keep this from being a full verification: (1) "surface figure" (low-order deviation from the ideal aspheric shape) and "MSFR" (mid-spatial-frequency roughness, closer to what "roughness" usually denotes in optics) are related but distinct metrics, both bundled together as "roughness" in the original claim; (2) the data are from 2011, describing the NXE:3300 generation — not necessarily the current state of the art (later High-NA systems are not covered by this source). The figure is therefore corroborated for a specific historical product generation, not shown to be an evergreen constant.

Claim: Independent, named Tier-2 analysts do characterize the ASML/Zeiss EUV advantage as resting on tacit, cross-organizational "know-how" rather than patents as the primary protective mechanism

Claim type: technical-mechanism. Source tier required: Tier 1-2. Source tier met: Tier 2 (Burja), supported at Tier 3 by an academic-institute framing (IP Business Academy).

Samo Burja's Bismarck Analysis writes that reliably reproducing ASML/Zeiss's photolithography capability "relies on intellectual dark matter: the tacit knowledge, idiosyncratic skill, and unwritten rules that are necessary for success but cannot be made legible to outsiders or non-experts except through training with the current expert practitioners, in whose heads the knowledge actually lives," and frames ASML/Zeiss as the container for a multi-generational "tradition of knowledge" traceable back through Philips's NatLab labs to the 1970s. A separate piece hosted by CEIPI's IP Business Academy makes a structurally similar argument via the Resource-Based View of the firm: EUV capability "sits in the interaction between ASML as system integrator, Carl Zeiss SMT as the provider of the optical system, TRUMPF..., Fraunhofer institutes and many other specialized suppliers," forming "a resource system" rather than a patent-protected artifact, one that "consists of routines, trust, supplier qualification, engineering memory, process knowledge, tacit expertise, investment discipline and the ability to coordinate many forms of knowledge across organizational boundaries" — explicitly arguing patents are one contributing mechanism among several, not the source of the advantage itself.

This clears the mechanism-claim floor at Tier 2: the "tacit-knowledge-not-patents" framing is a real, independently-argued analytical position among people who write under their own names about this industry, not an artifact invented by the original Tier-4 explainer site. It is best read as a claim about ASML/Zeiss's durability against future or hypothetical competitors (Burja's piece is explicitly framed around whether China could catch up) rather than a specific historical account of why Nikon and Canon in particular did not succeed — see the next claim.

Claim: The specific claim that Nikon and Canon spent "15+ years" failing to replicate ASML/Zeiss's tacit know-how is not supported by the better-sourced history — the documented reasons are political exclusion from the EUV-LLC consortium and a later strategic/financial exit, not a demonstrated technical failure to imitate

Claim type: historical (contested — corrects a specific causal mechanism). Source tier required: Tier 1-2 (escalated per the sourcing floor's contested-claim clause). Source tier met: Tier 2.

Brian Potter's Construction Physics account states that when the US-led EUV-LLC consortium formed in the mid-1990s, "Nikon declined to participate in EUV-LLC in part due to the resulting controversy, and Canon was ultimately prevented from joining by the US government," and concludes: "Left outside of the EUV-LLC consortium, Nikon and Canon never successfully developed EUV technology." Samo Burja's ASML piece adds the later chapter: "Nikon built a working prototype in 2008 but delayed and eventually abandoned its EUV program after the 2008 global financial crisis and Intel's 2012 decision to invest in ASML, signaling its commitment to buying their machines." Both accounts describe Nikon as having reached a working EUV prototype (2008) before exiting for financial and strategic-signaling reasons — a different story from "15+ years of failed attempts to enter," which implies persistent, unsuccessful technical effort rather than exclusion from the information-sharing consortium followed by a business decision to stop once the market had visibly consolidated around ASML.

This does not refute the tacit-knowledge thesis as a general account of why a new entrant would struggle today (that claim clears the floor above, on its own terms). But it means the original note's specific causal chain — Nikon/Canon tried for 15+ years and were blocked by tacit know-how specifically — conflates two different, better-documented mechanisms (consortium/political exclusion; later strategic exit) with a technical-imitation failure that neither Tier-2 source actually describes for Nikon or Canon by name.

Further leads

• Bryon Moyer, "Nikon vs. ASML" (EEJournal, 2017) documents an active Nikon patent suit against ASML/Zeiss over immersion lithography (DUV, not EUV) — a reminder that patents do matter for adjacent lithography technology even if they're argued not to be the load-bearing mechanism for EUV mirrors specifically; worth a dedicated capture on whether this complicates the "not IP" framing.
• The Zeiss 2011 slide deck's MSFR chart shows roughness dropping from roughly 102 picometers (NXE:3100 average, ~2009) to 52 picometers (first NXE:3300 mirrors, 2011) — a historical improvement curve worth its own quantitative claim if a primary source for the earlier (2004–2005, ~200 picometer) data point can be located.
• Whether current-generation High-NA EUV mirrors (post-2019) carry a tighter roughness/figure spec than the 2011 NXE:3300 numbers found here — not checked this session.
• Samo Burja, "Carl Zeiss' Tradition of Knowledge in Optics" (Bismarck Brief, 2025) is paywalled beyond its free preview; the full piece likely applies the tacit-knowledge thesis specifically to Zeiss's mirror-making rather than ASML generally and would be a stronger direct source if read in full.
• IP Business Academy's "resource system" / Resource-Based-View framing is unsigned ("Editorial staff") — worth checking whether CEIPI publishes named-author versions of the same argument elsewhere before leaning on it beyond Tier 3.

Entity candidates

• Hiroo Kinoshita — person — NTT researcher who first demonstrated reflective EUV/soft-X-ray lithography imaging (1985); the foundational originator of the technology ASML/Zeiss later commercialized, and the figure any "who really did EUV first" priority question should be measured against.
• Herman van Heek and Gijs Bouwhuis — persons — Philips NatLab engineers who built the first wafer-stepper prototype (1971–73), the direct institutional ancestor Burja traces ASML's "tradition of knowledge" back to, older than ASML itself.
• Tinsley (mirror-fabrication firm) — concept/company — built the Hubble Space Telescope's precision optics and supplied NTT's first multilayer EUV mirrors before Zeiss's role solidified; the older precision-optics lineage the modern Zeiss capability is a continuation of.
• Samo Burja — person — named analyst behind the two Bismarck Analysis pieces used as core Tier-2 mechanism sources.
• Brian Potter — person — named author of the Construction Physics historical account.
• Martin van den Brink — person — ASML co-president/CTO since the company's earliest years; identified by Burja as the living carrier of ASML's "tradition of knowledge."
• Carl Zeiss SMT GmbH — concept/company — the entity that actually holds the mirror-fabrication know-how at the center of the whole question.
• EUV-LLC — concept — the 1990s US consortium whose membership decisions (not tacit knowledge per se) determined who had early access to EUV research; worth its own note given how much causal weight both Tier-2 sources place on it.