1948 Bell Telephone Laboratories' Bardeen and Brattain Filed a 'Three-Electrode Circuit Element Utilizing Semiconductive Materials' Patent US2524035 — The Core Patent of the Point-Contact Transistor With Shockley Absent From the Inventor List, and the 1956 AT&T Consent Decree That Made It Royalty-Free as the Institutional Origin of Global Semiconductor Industry Diffusion

Conclusion First

On February 26, 1948, at Bell Telephone Laboratories in Murray Hill, New Jersey, physicist John Bardeen and experimentalist Walter H. Brattain filed US Serial No. 11,165 titled "three-electrode circuit element utilizing semiconductive materials." It was refiled as a continuation-in-part on June 17, 1948 (Serial No. 33,466) and granted on October 3, 1950 as US2524035. Claim 1 covers "a circuit element comprising a block of one conductivity type with a thin opposite-type surface layer, plus emitter, collector, and base electrodes" — this is the fundamental patent of the point-contact transistor.

William Shockley is absent from the inventor list of this patent. Shockley filed a separate junction transistor patent US2569347 (filed 1948-06-26, granted 1951-09-25, sole inventor). The two patents represent a design split — two different solutions to the same problem ("how to build a 3-terminal semiconductor amplifying element"). The 1956 Nobel Physics Prize was awarded jointly to all three, but the inventorship of this patent remained at two.

Then on January 24, 1956, the US Department of Justice vs. AT&T antitrust consent decree placed all pre-decree Bell Labs semiconductor patents under royalty-free compulsory licensing for all applicants (with RCA/GE/Westinghouse excluded due to existing cross-licenses). Texas Instruments, IBM, and Sony entered at low cost, and through them Toshiba, NEC, Samsung, and TSMC followed — making the consent decree the institutional origin of global semiconductor industry diffusion (Watzinger et al. 2020 rates this as "above Marshall Plan in economic contribution").

As the inaugural episode of Week 4 "Hardware & Energy Patent" subseries, this article re-reads this 77-year-old, 3-page-and-change patent document as the origin point of the contemporary China-AI × Korea-Taiwan semiconductor niche.

1. How the Subject Was Selected (Reproducible Pipeline)

[STEP 1] Extract un-published priority-16 candidates from candidates.tsv HW-001~010
         → HW-001 transistor patent (Bell Labs Bardeen-Brattain) chosen as the
           symbolic Week 4 opener
[STEP 2] Verify reachability of DB URL (https://patents.google.com/patent/US2524035)
[STEP 3] WebFetch Claim 1, inventors, assignee, dates from Google Patents
[STEP 4] Cross-check DB record "Bell Telephone Laboratories, Bardeen/Brattain,
         granted 1950-10-03" against primary source
         → All match (7th DB-match case in Day 8~15 19-correction series)
[STEP 5] Verify peripheral facts: Shockley absent inventorship, relation to
         junction patent US2569347, 1956 AT&T consent decree, joint Nobel 1956
[STEP 6] Connect to contemporary niche: trace Sony / Samsung / TSMC lineage
         through the institutional effect of the AT&T consent decree

Selection rationale: (a) maximum symbolic significance for Week 4 inaugural, (b) the historically important point of separated inventorship between Bardeen-Brattain and Shockley, (c) the 1956 AT&T consent decree as the institutional history that explains semiconductor globalization, (d) connecting Haruko's main niche (China-AI × Korea-Taiwan semiconductors) to its 77-year-old origin in a single session, (e) a rare DB-match case among the Day 8~15 19-correction series.

2. Claim 1 and the Core of the Specification

Claim 1 from Google Patents (verbatim):

A circuit element which comprises a block of semiconductive material of which the body is of one conductivity type and a thin surface layer is of the opposite conductivity type, an emitter electrode making contact with said layer, a collector electrode making contact with said layer disposed to collect current spreading from said emitter electrode, and a base electrode making contact with the body of the block.

Five key points:

1. The subject is a "circuit element." It claims the basic structure of a 3-terminal semiconductor amplifying element, not the entire device.
2. The semiconductor block has one conductivity type (n-type or p-type).
3. The surface layer is a thin opposite-type layer. This is the conceptual precursor to modern MOSFET inversion / channel layers.
4. Emitter and collector electrodes contact the surface layer. In point-contact design, metal needles directly touch.
5. The base electrode contacts the body.

Opening definition from the specification:

This invention relates to a novel method of and means for translating electrical variations for such purposes as amplification, wave generation, and the like.

The patent fixes the conception of replacing vacuum-tube amplification/oscillation with solid-state devices. Note that 1950 USPTO conventions had no required Abstract section (Abstracts became mandatory after 1976), so this patent has no separate Abstract. We treat the opening definition as Abstract-equivalent.

3. The Design Split Between Point-Contact and Junction — The Circumstances of Shockley's Absence

December 1947 Point-Contact Working Demo

Bell Labs' solid-state physics group was launched in 1945 by Shockley. The original concept was the field-effect amplifier (precursor to modern FET), led by Shockley but repeatedly failing in experiments. Bardeen proposed surface states theory in 1945, explaining how electron states at the semiconductor surface blocked the field effect.

In November–December 1947, Bardeen and Brattain repeatedly contacted metal needles to germanium (Ge) surfaces. On December 16, 1947, a device with two metal foil contacts spaced ~0.05mm apart on a Ge single crystal showed amplification. On December 23, 1947, the device demonstrated to Bell Labs management showed amplification gain of about 18×. This was the first transistor in history.

Shockley was not directly involved in this experiment. Per the Computer History Museum's account, Shockley felt strong dissatisfaction at being excluded from point-contact inventorship and from January–March 1948 independently developed the junction transistor concept.

Design of the Junction Transistor US2569347

Shockley filed US2569347 "Circuit Element Utilizing Semiconductive Material" as sole inventor on June 26, 1948, granted September 25, 1951. This is the core patent of the junction transistor.

Design split: (a) point-contact achieved the world-first lab demo but had low manufacturing reproducibility; (b) junction enabled mass production through crystal growth and became the standard form of the 1950s semiconductor industry. The two are "different structural solutions to the same problem (3-terminal semiconductor amplification)" — a precursor + parallel-invention relationship. The Bardeen-Brattain patent is not the direct ancestor of modern MOSFETs; the direct ancestor of modern MOSFETs is the 1959 Atalla-Kahng MOS structure patent.

4. The 1956 AT&T Consent Decree — The Institutional Origin of Semiconductor Industry Globalization

On January 24, 1956, the US DOJ and AT&T settled the antitrust case (United States v. Western Electric Co. & AT&T, filed 1949) via Consent Decree. Key terms:

1. Pre-decree patents (granted before decree): royalty-free compulsory license to all applicants (RCA / GE / Westinghouse excluded due to existing cross-licenses)
2. Post-decree patents: non-exclusive license at "reasonable royalties"
3. AT&T's business restricted to communication services; semiconductor manufacturing prohibited

US2524035 and US2569347 were both pre-decree and thus subject to royalty-free compulsory licensing. This enabled:

• Texas Instruments (1954 silicon transistor commercialization)
• IBM (mid-1950s, semiconductor memory research)
• Sony (1955 transistor radio TR-55, post-license mass production)
• Toshiba / NEC / Hitachi (late 1950s, AT&T-licensed semiconductor business launch)

to enter at low cost. The 2020 economic history paper by Watzinger, Fackler, Nagler, and Schnitzer ("Bell Labs and the 1956 Consent Decree") rates this consent decree's economic impact as "surpassing the Marshall Plan."

Subsequent lineage: 1960s Japanese semiconductor industry establishment → 1980s DRAM competition → 1990s onward Korea (Samsung, SK Hynix) and Taiwan (TSMC) rise → 2020s China-AI × Korea-Taiwan semiconductor roadmap. A 3-page patent of 77 years ago became the institutional origin of contemporary geopolitical semiconductor niches — this correspondence flows through the 1956 consent decree's "license compulsion" relay point.

5. Why the Comparison Feels "Uncannily Close" (Correspondence Table)

No row reaches "identity" level. Saying point-contact transistors are "the same as" modern MOSFETs is a design-level error — modern MOSFETs descend from the 1959 Atalla-Kahng patent US3102230, with 3D FinFETs from Hisamoto et al. 1989 (JPS62-105474A), forming a separate lineage. This patent's significance is its position as a precursor that first solved the 3-terminal semiconductor amplification problem; subsequent MOSFET/FinFET/GAA lineages all sit atop this problem framing.

The correct phrasing is: "Designs differ but problem framings overlap."

6. Why It Was Forgotten (Speculation)

The patent itself is not forgotten. It's referenced in every general history book, textbook, and Computer History Museum exhibit on Bell Labs. What's forgotten is rather "what the patent text (Claim 1 verbatim) covers as the origin point of contemporary semiconductor industry":

1. Almost no one reads Claim 1. History books and textbooks describe the December 23, 1947 demo, but rarely quote the patent application document itself.
2. The institutional effect of the 1956 consent decree is rarely cited in popular discourse. The "Bell Labs invented the transistor" hero narrative circulates, but the institutional history side ("global diffusion happened because of the consent decree's royalty-free terms") is confined to economic history specialist papers.
3. Shockley's absence from inventorship is rarely surfaced outside tech-history nerds. The discrepancy ("Nobel Prize jointly to 3, but patent stays at 2") is only verifiable from primary sources.

7. Significance for AI Archaeology

For the series problem framing — "re-reading the long documents humans didn't read, with LLMs" — this patent is a typical excavation target:

• The patent itself is just over 3 pages, not "long," but almost no readers actually read Claim 1 verbatim — qualifying as an unread long document.
• The 25-page Consent Decree is a representative example of an unread document outside economic history specialists (we did not access the full text in this episode; future Note episodes may dig deeper).
• Re-positioning the contemporary China-AI × Korea-Taiwan semiconductor niche on top of "a 3-page patent + 25-page consent decree from 77 years ago" reveals the institutional pre-history of geopolitical semiconductor hegemony.

LLMs are optimal for fast-scanning this kind of "short but unread" primary source set.

8. Pitfalls Specific to Hardware Archaeology

Pitfall 1: Don't write "direct ancestor." This patent is not the "direct ancestor" of modern MOSFETs. The MOSFET lineage descends from the 1959 Atalla-Kahng patent US3102230, designed differently. We consistently write "important precursor" / "prior example."

Pitfall 2: The Nobel-3 / Patent-2 discrepancy. The 1956 Nobel Physics Prize was joint among Shockley/Bardeen/Brattain, but US2524035 inventors are only Bardeen/Brattain. Shockley is the sole inventor of US2569347. Do not conflate "Nobel Prize = patent inventorship."

Pitfall 3: Scope of the 1956 consent decree. The decree's scope is limited to Bell Labs (AT&T subsidiary) patents. Contemporary semiconductor patents from RCA/GE/Philco etc. were outside its scope. Generalizing to "the entire semiconductor industry became royalty-free" is incorrect.

Pitfall 4: Overuse of contemporary-niche analogies. "US-China semiconductor regulation as inverted structure of 1956 consent decree" is an analogy, not a structural identity. The former is export-control license-restriction, the latter antitrust license-compulsion — opposed structures, working as analogy but not as analytical model.

Strictly Speaking

Confirmed facts:

• Retrieved from Google Patents: US2524035 Claim 1, inventors (Bardeen / Brattain, 2), assignee (Bell Telephone Laboratories Inc.; current AT&T Corp), priority (1948-02-26 from Serial No. 11,165 abandoned parent, CIP), filing (1948-06-17), grant (1950-10-03), expiration status (Expired Lifetime)
• Confirmed Shockley junction patent US2569347 (filed 1948-06-26, granted 1951-09-25, sole inventor) via secondary Google Patents fetch
• 1956 January 24 AT&T Consent Decree confirmed via Watzinger et al. 2020 paper and PBS Transistor program article
• 1956 Nobel Physics Prize joint award to Shockley/Bardeen/Brattain confirmed via NobelPrize.org

Author's interpretation:

• "The institutional origin of the contemporary China-AI × Korea-Taiwan semiconductor niche is the 1956 consent decree" is interpretation, not validated by quantitative economic models (Watzinger et al. is the closest empirical work)
• "Point-contact and junction are two structural solutions to the same problem" is a tech-history synthesis; the patent texts themselves do not perform this comparison

Analogies:

• "US-China semiconductor regulation = inverted structure of 1956 consent decree" is analogy
• "China RISC-V vs. ARM vs. x86 ISA split = structurally similar to point-contact vs. junction split" is conceptual analogy, design-level different
• "Shockley's absence ≈ Jensen Huang / Lisa Su symbolic representation" is analogy, structurally different

Unconfirmed:

• Full text (25+ pages) of the 1956 Consent Decree was not read in this episode; citations are via secondary sources
• Bell Labs internal license negotiation records (AT&T Archive at Murray Hill) not retrieved
• Forward citations count not retrievable due to Google Patents display limits
• Whether TSMC / Samsung's early licenses were obtained via AT&T (only secondary-source confirmation via Watzinger et al.)

Where this comparison breaks:

• "77-year-old patent = origin of contemporary semiconductor niche" is institutional-history synthesis. The technical lineage flows through many relay points: Atalla-Kahng MOS 1959, Dennard scaling 1974, Mead-Conway VLSI 1980, Hoeneisen-Mead 1972 etc. — relay points this article skips.
• The 1956 consent decree's effects extend beyond semiconductors to lasers, communication, computing as a whole, so isolating its semiconductor-only effect is post-hoc synthesis.

References:

• Google Patents: US2524035
• Google Patents: US2569347 (Shockley junction)
• Computer History Museum: 1947 Point-Contact Transistor
• Computer History Museum: 1948 Junction Transistor
• NobelPrize.org: Physics 1956
• PBS Transistor: Deciding to Share (1956 Consent Decree)
• Watzinger, Fackler, Nagler, Schnitzer 2020: Bell Labs and the 1956 Consent Decree