US 4728591

I'll search for this specific patent number and check for any litigation docket activity.

Let me check specifically for litigation or CAFC activity involving this patent.

Search scope and caveats

I searched for the literal string 4728591 across USPTO/Google Patents, Justia, EPO/Espacenet, uspto.report, and general web results, plus 2026 CAFC docket reporting. Two important literal-number caveats before the summary:

1. A conflicting document shares the same numeric string. A Japanese publication JP 4728591 A (published 1991-02-21) appears in EPO Global Patent Index family records for EP 0500364 ("Optically active 1,3-dioxin-4-ones," Chisso). That is a different document on a different subject and is not the US patent at issue. I am reporting it literally rather than treating it as the same patent.
2. No CAFC 2026 docket activity for US 4,728,591 was found. My searches returned extensive 2026 CAFC material (e.g., AudienceScience v. Google, Teva v. Lilly, International Medical Devices v. Cornell) and historical Boston University LED litigation (the '738/'703 GaN patents, Case Nos. 1:12-cv-11935, 1:13-cv-10659, etc.), but nothing tying US 4,728,591 to any Federal Circuit or district court docket. I cannot confirm this is exhaustive — I do not have direct PACER/CAFC docket access, so treat this as "not found," not "confirmed none."

---

US Patent 4,728,591 — summary

Assignments (from Google Patents reassignment records):

• 1987-10-01 — assigned to University of Colorado Foundation, Inc. (assignors: Rothschild, Douglas, Clark)
• 1987-11-27 — assigned to Trustees of Boston University (assignors: Rothschild, Douglas, Clark)

Abstract (as published)

Articles exhibiting fabricated structures with nanometer size scale features (nanostructures), typically a device comprising nanostructures of a functional material on or in a substrate of dissimilar material, are produced by a method employing a substrate base or coating and a thin layer serving as a lithographic mask or template, consisting of a self-assembled ordered material array, typically a periodic array of molecules such as undenatured proteins, exhibiting holes, thickness or density variations. It is possible to produce complex structures containing large numbers of nanometers scale elements through a small number of simple steps.

Plain-language overview

The patent's core idea is to stop writing nanoscale patterns serially (focused e-beam) and instead borrow a pattern nature already made. A self-assembled, two-dimensionally ordered molecular array — typically a crystallized, non-denatured protein such as the S-layer of Sulfolobus acidocaldarius, purple membrane from Halobacterium halobium, rhodopsin, hemocyanin, cytochrome oxidase, or porin — is bound to a substrate. These arrays have 1–50 nm periodicity and contain ordered holes, thickness/density variations, or even chemical-reactivity variations. That array is then used three ways:

1. As a template for an overlayer — deposit a thin film (1–2.5 nm, e.g., Ta/W) on top and ion-mill/X-ray/etch through the array's holes to perforate the film, producing a nanometer "screen."
2. As a mask for deposition — evaporate material through the array's holes onto the substrate, then strip the biological array (detergent, protease, acid) leaving a periodic array of nanoscale islands.
3. As a mask for etching — use the holes to pattern (pit) the substrate beneath.

The payload is parallelism: thousands of identical nanometer elements are produced in a few steps instead of one at a time. The specification extends this to multilayered substrates (Si/SiO₂, MBE-grown heterostructures), 3-D patterning, and biological/solid-state hybrids — biosensors with bacteriorhodopsin in the holes, SIS tunnel-junction arrays, flux-pinning superconducting screens, 2-D/3-D superlattices, and even a bR-based molecular memory addressed by an STM probe.

Independent claims (plain language)

From the claim set as retrieved, the independent claims are 1, 2, 3, 5, 6, and 7 (claim 4 depends from claim 3). Note: the patent has 38 total claims; my retrieved listing covers the apparatus claims but not the method claims (roughly claims 23–38) — I flag that as a gap rather than guess at their text.

• Claim 1 — Nanosubstrate with a patterned overlayer and electrode. A substrate base; a self-assembled 2-D array on it forming a spatial pattern of thickness, density, and/or chemical-reactivity variation with a 1–50 nm characteristic dimension; a patterned layer on the array whose pattern is determined by the array's pattern; and at least one electrode coupled to the patterned layer or the array. (The "determined by" and "electrode" limitations are what make it a device, not just a coated slide.)

• Claim 2 — Projection/discrete-island article. A base with a plurality of projections in a 1–50 nm pattern corresponding to the thickness/density/reactivity pattern of a self-assembled 2-D array; plus a layer of a different material disposed on the base between the projections, where that material does not naturally form the pattern. (This is the "fill the gaps with a second material" — e.g., insulator between metal islands — structure of FIG. 7D.)

• Claim 3 — Nanosubstrate with state-determination means. Substrate base + self-assembled 2-D array with a 1–50 nm pattern, plus means for determining the state of selected portions of the array. (A functional/means-plus-function claim aimed at readout/addressing of the patterned array.)

• Claim 4 (dependent on 3) — the determining means is a plurality of electrodes coupled to the array.

• Claim 5 — Patterned base with depressions. A substrate base + self-assembled 2-D array forming a thickness/density/hole pattern of 1–50 nm, where the substrate base has depressions corresponding to the array's holes. (The etched-pit embodiment — FIG. 3 / FIG. 8B.)

• Claim 6 — Overlayer substantially only on the array. A substrate base + self-assembled 2-D array + a patterned layer disposed on the array and substantially not on the substrate base, having a pattern of holes determined by the array's pattern. (Masks the "lift-off"/selective deposition variant.)

• Claim 7 — Overlayer with holes (broader). Substrate base + self-assembled 2-D array + patterned layer on the array having a pattern of holes determined by the array's pattern — without the "substantially not on said substrate base" limitation of claim 6, making it the broader perforated-screen claim.

Dependent claims of note: claim 8 restricts the array to spatially ordered nondenatured proteins; claims 9–14 name specific biologics (S-layer of Sulfolobus acidocaldarius, purple membrane of Halobacterium halobium, rhodopsin, cytochrome oxidase, E. coli matrix porin, tight gap junctions); claim 15 covers a 3-D crystal surface layer (claim 16: protein crystal); claim 17 covers ordered holes of 1–50 nm; claim 18 covers chemical-reactivity-variation patterns; claim 19 covers multilayer substrate bases; claims 20–21 cover alignment (commensurate) vs. non-alignment (incommensurate) of the array pattern with substrate heterogeneity.

Uncertainty notes

• Claim list completeness: I have independent and many dependent apparatus claims via a secondary aggregator (Justia's claim listing). The patent states 38 claims; method claims are not fully represented above.
• Legal status: "Expired – Fee Related" with anticipated expiration 2006-03-07 is Google Patents' own disclaimer-laden characterization; I have not verified it against a USPTO fee/expiration record directly.
• Assignee history: The 1987 reassignment entries show assignments to both University of Colorado Foundation and Trustees of Boston University from the same three inventors — consistent with the co-original-assignee listing, but the reason for the dual assignment is not something I can establish from the entries alone.
• CAFC 2026: No docket match found; this is a negative search result, not an affirmative confirmation of no litigation.

Sources: https://patents.google.com/patent/US4728591/en; https://patents.justia.com/patent/4728591; https://uspto.report/patent/grant/4728591; https://patents.google.com/patent/US4802951/en; http://data.epo.org/gpi/EP0500364A1 (for the unrelated JP 4728591 A).

I'll research litigation involving US Patent 4,728,591. Let me run several targeted searches.

Let me search additional litigation databases with more specific queries.

Known Litigation Involving U.S. Patent No. 4,728,591

Summary: I found no litigation involving U.S. Patent No. 4,728,591. Based on the searches I was able to run, there is no record of any district court, ITC, Federal Circuit, or PTAB proceeding in which this patent was asserted, challenged, or otherwise at issue.

The patent itself (for identification discipline)

Before reporting the negative result, it's worth confirming exactly which patent we're discussing, because "591" is heavily overloaded in the results:

Because the patent expired in 2006, any assertion window is 1988–2006.

What the searches actually returned — and why none of it is this patent

Every "591" hit I found refers to a different patent. I flag these explicitly so they are not mistakenly attributed to 4,728,591:

• Demodulation, Inc. v. United States (U.S. Court of Federal Claims, No. 11-XXXX; opinion at 111 Fed. Cl. 674) — concerns U.S. Patent No. 6,270,591, a magnetostrictive/sensor patent licensed from the Romanian National Institute. Different patent, different technology.
• DataQuill Ltd. v. High Tech Computer Corp. (HTC) (S.D. Cal.; IPRs) — concerns DataQuill's '591 patent on handheld-device data entry, plus a separate 40 patents. Different patent.
• Hydro / Petter litigation (W.D. Mich., 1:07-cv-01033) — the asserted '591 patent there is a vehicle-wash-pad patent. Different patent.
• Edge Systems LLC v. Aguila — '591 Patent relating to skin-treatment instruments. Different patent.
• ZimVie / Mobi-C cervical-disc dispute (PTAB petition, IPR) — '591 patent on artificial disc systems. Different patent.
• Design Patent D971,591 S (N.D. Ill., counterclaim, July 2024) — baby-crib-backpack design patent. Different patent (and a design patent, not a utility patent).
• Leigh Rothschild / RBDS / Valve Corporation v. Rothschild (W.D. Wash., 2:23-cv-01016, jury verdict Feb. 2026) — concerns US 8,856,221 B2 and other Rothschild patents; "Rothschild" here is Leigh Rothschild, not inventor Kenneth J. Rothschild of the '591 patent. Unrelated.
• JP 4728591 (Japanese voltage converter) and JP 4728591 A (Chisso chemical application) — unrelated foreign publications that coincidentally share the number.
• A Florida LLC annual report FEI No. 93-4728591 and a Brazilian business registry entry "4728591" — administrative coincidences, not patents.

What 4,728,591 does show up in (not litigation)

The patent appears extensively in the technical literature, which is easy to confuse with legal activity:

• Cited as prior art / background in later patents and applications, e.g., U.S. Pat. No. 6,890,624 ("incorporat[ing] U.S. Pat. No. 4,728,591 to Clark et al. . . . by reference"), U.S. Pat. No. 6,998,228, EP 1,081,513, and WO 2002/033461 (where it is an "X" category reference).
• Cited in an EPO opposition against EP 1,086,742 B2 (Affymax combinatorial chemistry).
• Subject of bibliometric/citation studies (e.g., the NSF-funded nanotechnology patent analysis published in J. Nanoparticle Research, DOI 10.1007/s11051-005-5468-3, which ranks 4728591 first by Authority Score in USPC classes 428 and 430, crediting "Trustees of Boston University").
• Google Patents shows no "Litigation" or "Reexamination" entries in the patent's event timeline — only the 1986 filing, the 1987 assignments of interest, the 1988 grant, and the 2006 expiration.

Caveats you should weigh

I want to be candid about the limits of this result, since you're asking a due-diligence question:

1. This is a negative finding from web search, not a certified PACER/Docket Navigator/PACER-wide database report. I was not able to run a full Docket Navigator, Lex Machina, or comprehensive PACER docket-number search within my tool limits. A conclusion of "no litigation ever" in a strict legal sense should be verified with one of those paid sources.
2. I could not successfully retrieve the Unified Patents litigation case list (portal.unifiedpatents.com/litigation/caselist) or a Docket Alarm / RPX Insight party-page for this patent number directly. Unified's own published IPR-filing tables (2013–2021) list no 4,728,591 challenge.
3. The searches above are consistent with — but do not by themselves prove — the proposition that the patent was never asserted. Several factors make a zero-litigation history plausible: it was university-owned (Boston University / University of Colorado Foundation), it expired for failure to pay maintenance fees rather than being maintained for licensing value, and there is no recorded assignment to a monetization entity.

Bottom line

No litigation is known to me involving U.S. Patent No. 4,728,591 (Clark, Douglas & Rothschild; Boston University / University of Colorado Foundation). There is no plaintiff, defendant, jurisdiction, case number, filing date, or outcome to report, because I found no case. The numerous "591 patent" suits that surface in litigation databases are directed to seven- and eight-digit patent numbers (6,270,591; 8,856,221-derived families; D971,591; etc.) that are legally and technically distinct from the five-digit 4,728,591, and those should not be attributed to it. If you need a court-certified negative, I'd recommend a Docket Navigator or Lex Machina "patent at issue" report run against 4,728,591 across all federal districts plus the ITC and PTAB.

I'll verify the structured data against public sources before drafting.

Let me run a couple more targeted checks before concluding.

One more check on the related continuation patent, to give a complete family picture.

Proceedings overview

Total AIA trial proceedings on file for US 4,728,591: zero (0). The structured "PTAB proceedings on file" block — the USPTO ODP canonical list — returns no IPRs, no PGRs, and no CBM reviews, so the breakdown is: active 0 / claims invalidated 0 / claims sustained 0 / settled 0 / institution denied 0. My independent web checks (Google Patents, Justia, PTAB-decision aggregators, general docket searching) surfaced no proceeding either, and no Federal Circuit appeal with this patent in the chain. The defensive posture this gives a defendant is not "hardened patent" but "dead patent": the patent's own record shows an anticipated expiration of 2006-03-07, which is before the first IPR petition could legally have been filed (2012-09-16). The absence of PTAB activity here is not a litigation signal at all — it is a calendar fact. Anyone receiving a demand letter citing US 4,728,591 today should be asking a much more basic question than "what did the Board hold," namely "how can this patent still be in force."

⚠️ Identifiers to watch for — do not auto-correct these. Searches for "4728591" return several unrelated false positives that are not this patent: JP 4728591 A / JP 4728591 B2 (Japanese patent numbers — a voltage converter and an optically active 1,3-dioxin-4-one), an FEI/EIN corporate registration number "93-4728591," and a color value 0.4728591 inside a Collada 3D-format file. None has any relationship to US 4,728,591.

---

No proceedings to report

There is no {PROCEEDING_NUMBER} to list. I will not invent one, and there is no petitioner, no APJ panel, no institution decision, no Final Written Decision, and no settlement to describe. Every field the task template asks for is empty because the underlying proceeding does not exist.

For completeness, the record facts that are documented:

• Patent: US 4,728,591 A — "Self-assembled nanometer lithographic masks and templates and method for parallel fabrication of nanometer scale multi-device structures"
• Application: US 06/837,376
• Inventors: Noel A. Clark, Kenneth Douglas, Kenneth J. Rothschild
• Assignees (original): Boston University; University of Colorado Foundation, Inc.
• Priority / filing date: 1986-03-07
• Grant date: 1988-03-01
• Anticipated expiration (per Google Patents): 2006-03-07
• Legal status: "Expired - Fee Related"
• Source: https://patents.google.com/patent/US4728591/en

Expiry arithmetic (my own, checkable): this is a pre-URAA patent that was in force on 1995-06-08, so the URAA transitional rule gives the greater of 17 years from issue (1988-03-01 → 2005-03-01) or 20 years from filing (1986-03-07 → 2006-03-07). The greater is 2006-03-07, which matches the record. One honest caveat: the "Expired - Fee Related" label is normally the hallmark of a maintenance-fee lapse, which could mean the patent actually died earlier than 2006-03-07. I cannot resolve that from the data available to me — but it does not matter, because either way the patent was gone years before AIA trials existed.

Related continuation, for the family picture: US 4,802,951 A ("Method for parallel fabrication of nanometer scale multi-device structures"), application 07/136,992, granted 1989-02-07, same Boston University assignee lineage. No PTAB activity surfaced for it either, and it too is long expired. Source: https://patents.google.com/patent/[US4802951A](/patent/US4802951A)/en

---

Strategic summary

Claim status: UNTESTED — and moot. No claim of 4,728,591 has ever been canceled, confirmed, or even construed by the PTAB, because no petition was ever filed. That is worse for a patent owner than an adverse FWD, not better: it means there is no institutional validation of the claims whatsoever. The claims were never challenged at the Board, and — per the expiration date above — the statutory window in which they could have been challenged never overlapped with their enforceable life. The threshold point is that a patent whose anticipated expiration is 2006-03-07 cannot support an infringement claim in 2026. Damages would be limited to the pre-expiration period at best, and the six-year limitation of 35 U.S.C. § 286 reaches back from today only to 2020-09-27 — a period that postdates expiration by roughly fourteen years. An assertion today would, on these facts, be worth nothing as a damages matter and is a strong candidate for Rule 11 / § 285 treatment if pressed.

Estoppel landscape: N/A, and the reason is structural. Section 315(e)(2) estoppel only binds a party that was a petitioner, real party in interest, or privy in an IPR that reached a final written decision. With zero proceedings, there is no estoppel on any prior-art ground, no § 325(d) discretionary-denial baggage, and no claim-construction record to inherit. But this cuts both ways: it also means a defendant has no ready-made invalidity record to borrow and no adjudicated cancellation to point at. The invalidity case would have to be built from scratch — and given the 1986 filing date and the dense, well-documented 1980s nanolithography literature cited in the specification itself (Carter on molecular-electronics fabrication, Craighead on 20-nm-scale e-beam metal particle arrays, Mochel), such a case against a 1986 priority date would be unusually strong. In practice none of that is necessary: expiration ends the analysis before validity is reached.

Pattern signals: none — and this is the tell. No serial petitioner, no defensive aggregator (no Unified Patents or RPX involvement anywhere in the chain), no patent-owner appeal campaign, no reissue or reexamination activity that I could find. The only documented enforcement posture by Boston University in this technology space came much later and involved an entirely different patent family — the Moustakas GaN thin-film patents (filed 1991, granted 1997) asserted against Apple, Amazon, Samsung, NEC, and roughly forty other defendants beginning in 2013 (e.g., Trustees of Boston University v. NEC Corp., 1:13-cv-12340-PBS (D. Mass.), and the coordinated Massachusetts cases before Chief Judge Patti B. Saris). I found no evidence that US 4,728,591 itself was ever asserted in litigation. Caveat: I did not run an exhaustive district-court docket search across all 94 districts, so I can rule out PTAB activity with high confidence and litigation activity only with moderate confidence.

The general truism in the task framing — "well-asserted patents eventually attract IPRs" — is worth refining here. US 4,728,591 is a highly cited patent (it ranks first by "authority score" in its USPTO class in the NSF-funded nanotechnology patent study, and is cited by later nanotechnology patents such as the Winningham/Douglas 2002 filings). Heavy citation, however, is a signal of technological importance, not of assertion importance. A 1986 patent that expired in 2006 is cited as foundational background art, never as a live infringement target.

---

Recommended next steps

If you are a defendant or a recipient of a demand letter citing US 4,728,591:

1. Lead with expiration, not invalidity. The patent's anticipated expiration is 2006-03-07 (see https://patents.google.com/patent/US4728591/en). For anything resembling current conduct, there is no actionable infringement period inside the § 286 six-year damages window (which reaches back only to 2020-09-27). Put the expiration date in your first response letter and demand the plaintiff identify the allegedly infringing acts and their dates.
2. There is no FWD to link to. I want to be explicit here rather than pad the record: because no proceeding exists, I cannot link you to an institution decision, a Final Written Decision, or a claim-level disposition — there is no such document. Do not let anyone hand you a "PTAB decision on 4,728,591"; it would be fabricated.
3. Verify the maintenance-fee history directly. The "Expired - Fee Related" status flag may mean the patent lapsed for non-payment before 2006-03-07. Pull the fee record on USPTO Patent Center (https://patentcenter.uspto.gov/) for application 06/837,376. An earlier lapse date only strengthens your position.
4. Check the continuation independently. US 4,802,951 A is a separate patent with a separate term; if a demand letter bundles it in, confirm its own expiration and fee history rather than assuming the 4,728,591 analysis transfers.
5. If a plaintiff persists after expiration is raised, preserve the correspondence. Asserting an expired patent against post-expiration conduct, with knowledge of the expiration date, is a strong § 285 exceptional-case argument and potentially sanctionable.

If you are the patent owner or its licensee: no IPR was ever filed against this patent, so there is no estoppel or adverse claim construction to work around — but that is cold comfort, because the patent is expired and the claims cannot be revived. There is no PTAB path forward (PGR was never available for a 1986 filing; CBM was limited to financial-services business methods and sunset on 2020-09-16; IPR could only have been filed from 2012-09-16 onward, six-plus years after the patent died). Any residual value in this family resides in later continuations and in the Boston University GaN portfolio, not here.

Sources consulted: Google Patents record for US 4,728,591 (https://patents.google.com/patent/US4728591/en) and US 4,802,951 A (https://patents.google.com/patent/US4802951A/en); USPTO PTAB E2E (https://ptab.uspto.gov/) and the USPTO PTAB Decisions portal (https://developer.uspto.gov/ptab-web/#/search/decisions); CourtListener (https://www.courtlistener.com/). No PTAB or Federal Circuit document specific to this patent exists to cite.

I'll research this patent's assignment and litigation history before drafting.

Let me dig into litigation/assertion history and the inventors' affiliations.

Let me check for any post-issuance transfers, litigation, or NPE activity tied to this patent family.

Ownership Reconstruction — US 4,728,591

Analyst note on source limits (read first). I could not complete a live query of USPTO Assignment Center in this session; both assignmentcenter.uspto.gov and assignment.uspto.gov are JavaScript single-page applications whose result grids are not returned to a text fetch, and Google Patents' "Legal Events" panel — the only machine-readable mirror I could reach — truncates each recording to date / assignee / assignor names / conveyance label. It publishes no reel/frame numbers, no execution dates, and no correspondent of record for this patent. Reel/frame and correspondent fields below are therefore marked "not exposed by indexed sources — verify at Assignment Center" rather than filled with placeholders. I have not invented any reel, frame, or attorney name.

---

Inventors

Pattern observations:

• The two co-assignees map cleanly onto the inventors' employers (CU inventors → University of Colorado Foundation; BU inventor → Trustees of Boston University). This is the ordinary university-sponsored-research pattern, not a fire-sale signature.
• Not present: the "all inventors departing the original assignee within 12 months of filing" tell. All three remained in academia; Clark is still a CU Boulder professor, Rothschild built the BU Photonics Center. No exodus preceded a sale.
• Diligence note (not an NPE signal): the application was filed 1986-03-07, three days before the enabling companion paper published on 1986-03-10. Had the paper published first, the § 102(a)/(b) posture of a 1986 filing would have needed attention. This is a validity/priority observation, not an ownership one.
• Dual-ownership split is the structurally unusual feature: from issuance the patent was held in undivided co-ownership by two unrelated nonprofit institutions. A two-owner patent is a standing/joinder wrinkle for any future assertion, and neither owner has recorded a transfer.

---

Original assignee

Two co-assignees on the face of the issued patent:

1. Trustees of Boston University — 147 Bay State Road, Boston, Massachusetts 02215, a Massachusetts corporation (exact address as recorded in the assignment paper). Nonprofit research university; primary business is education and research, with patent monetization run through its technology-transfer office. Status: operating. No product embodying the claims is known — the '591 claims are directed to a fabrication method (self-assembled protein monolayer as a nanometer lithographic mask/template), not a commercial article. I have no evidence BU ever licensed or practiced the '591 method commercially.

2. University of Colorado Foundation, Inc. — nonprofit foundation that historically held patent rights on behalf of the University of Colorado system (CU's tech-transfer function is now branded Venture Partners at CU Boulder). Status: operating as a nonprofit IP-holding affiliate. Again, no product embodying the claims identified.

BU is a litigious university — but on different patents. Its well-known assertion campaign ran on U.S. 5,686,738 and 6,953,703 (Moustakas GaN/LED patents) in Trustees of Boston University v. Epistar/Everlight/Lite-On, 1:12-cv-11935 (D. Mass.), the 2013 wave of LED/electronics suits, and the Samsung suit, 1:13-cv-10659 (D. Mass.). Those are university-own-IP suits against actual manufacturers. None of the reported pleadings, IPR papers, or judgments name U.S. 4,728,591. Treat BU's general enforcement posture as background, not as evidence about this patent.

Current legal status: "Expired – Fee Related"; anticipated expiration 2006-03-07 (20-year limit from the 1986-03-07 filing, which is later than 17-years-from-grant, 2005-03-01). The patent is dead; Google's "Current Assignee" field still shows Trustees of Boston University and University of Colorado Foundation Inc, i.e., no post-expiration transfer ever recorded.

---

Assignment timeline

Two recorded assignments exist, both pre-issuance, both non-adverse (inventors → their employers). Everything below is taken from the Google Patents legal-events panel; execution dates and reel/frame are not exposed there.

• Execution date not exposed / recorded 1987-10-01 — Reel not exposed by indexed sources — verify at Assignment Center

  • Conveyance: Assignment of Assignors' Interest (the "Assigned to" event carries the literal label "ASSIGNMENT OF ASSIGNORS INTEREST")
  • Assignor: Rothschild, Kenneth J.; Douglas, Kenneth; Clark, Noel A.
  • Assignee: University of Colorado Foundation, Inc., The
  • Correspondent: not exposed by indexed sources
  • Context: Original employer assignment — the CU-affiliated inventors perfecting title in their university's nonprofit IP-holding foundation. Not a sale; no consideration-driven transfer to a third party.

• Execution date not exposed / recorded 1987-11-27 — Reel not exposed by indexed sources — verify at Assignment Center

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Rothschild, Kenneth J.; Douglas, Kenneth; Clark, Noel A.
  • Assignee: Trustees of Boston University, 147 Bay State Road, Boston, Massachusetts 02215, a corp. of Massachusetts
  • Correspondent: not exposed by indexed sources
  • Context: Original employer assignment — BU perfecting title in the Boston University inventor's share; creates the two-owner co-ownership structure.

• No further assignments of record. The related continuation US 4,802,951 (application 07/136,992, filed 1987-12-23, claiming priority to this case and issued 1989-02-07) is a family member, not an assignment event — Google's event list can be misread as a transfer because it appears as "Priority to US07/136,992 / patent/US4802951A". It is not a conveyance. Any complete chain for this family must be reconstructed at the family level, since the '951 shares inventors and assignees.

If Assignment Center in fact returns no additional rows beyond these two 1987 recordings — which is what the indexed record indicates — that is itself the finding: the two original university assignees have owned the patent continuously from pre-issuance to expiration. I could not execute the definitive query, so treat the "no further rows" statement as the indexed record's conclusion pending your own Assignment Center run.

---

Timeline diagram

timeline
    title Ownership of US 4728591
    1986 : Application filed 07 Mar
         : Inventors Clark Douglas Rothschild
    1987 : Assigned to University of Colorado Foundation
         : Assigned to Trustees of Boston University
    1988 : Patent issued 01 Mar
    1989 : Continuation US4802951 issued
    2006 : Patent expired 07 Mar

---

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT. No assignee in the chain carries an "IP / Patents / Licensing / Holdings / Ventures" suffix, and no LLC appears at all. Both recorded assignees are the original nonprofit university entities (1987-10-01 and 1987-11-27 recordings). No registered-agent address, no single-purpose Dana LLC. Chain terminates at the co-owners.

2. Known asserter in the chain — NOT PRESENT. No assignee matches Acacia, Marathon, IV, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, or any Spangenberg entity. The Trustees of Boston University is a university, not a listed high-frequency NPE plaintiff; its documented campaigns (1:12-cv-11935, 1:13-cv-10659, the 2013 LED/electronics wave) were on the '738 and '703 patents, not on the '591.

3. Repeat correspondent across the chain — UNCLEAR. The correspondent of record is not exposed in any indexed source I could reach, so recurrence cannot be measured. This is the one signal that genuinely requires the Assignment Center detail view (correspondent + reel/frame) to close out. With only two recordings by two established university assignees, and no third-party assignee, the prior probability of a repeat-player NPE filing agent is very low — but mark it unknown, not absent.

4. Cascading transfers — NOT PRESENT. Two recordings, both in 1987, both inventor→employer, ten weeks apart, no chained LLCs, no common principals across adverse parties.

5. Pre-litigation transfer — NOT PRESENT. No infringement suit naming U.S. 4,728,591 was surfaced in any search (Litigation, CourtListener, Law360, IPR papers). A pre-litigation transfer requires a litigation; there is none.

6. Bankruptcy fire-sale — NOT PRESENT. Neither assignee is a debtor in insolvency; no Chapter 7/11 sale of this patent appears.

7. Privateering — NOT PRESENT. No operating-company-to-NPE transfer exists to be characterized; the co-owners never divested.

8. Defensive aggregator — NOT PRESENT. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. The inverse conclusion nonetheless holds: the asset was not weaponized, and it has been dead since 2006-03-07, so it is de facto neutralized as of today.

---

Verdict

Insufficient data — the chain contains only the original pre-issuance assignment recordings to the two university co-owners (University of Colorado Foundation, recorded 1987-10-01; Trustees of Boston University, recorded 1987-11-27), with no post-issuance conveyance of any kind, which under the stated rule means the original assignees still owned the patent through its 2006-03-07 expiry. No NPE signal is present on the available record, and the operative caveat is evidentiary rather than substantive: reel/frame numbers, execution dates, and the correspondent of record were not retrievable from indexed mirrors, so signals 1 and 3 cannot be affirmatively closed from a secondary source. Substantively this is a non-asserting dual-university ownership chain (BU plus the CU foundation), with BU's litigiousness confined to unrelated GaN/LED patents — not an NPE chain.

Verify at: https://assignmentcenter.uspto.gov/ (search by patent number 4728591; also run the continuation US 4,802,951) and the parallel index at https://assignment.uspto.gov/patent/index.html — pull the reel/frame, execution date, and correspondent for both 1987 recordings to close signals 1 and 3.

I'll search for the specific patent and its cited references.

Let me get the full citation list from the patent's "References Cited" section.

Let me retrieve the citation tables from Justia and Google Patents directly.

Prior-art analysis for US 4,728,591

0. Document identification (literal, no auto-correction)

The search returns US 4,728,591 A — Self-assembled nanometer lithographic masks and templates and method for parallel fabrication of nanometer scale multi-device structures, Appl. No. 06/837,376, filed 1986‑03‑07, granted 1988‑03‑01, Clark / Douglas / Rothschild, Boston University + University of Colorado Foundation. This is the document analyzed.

Numbers I am explicitly not treating as US 4,728,591 (they surfaced in the same search space):

• JP 4728591 A (pub. 1991‑02‑21) — Chisso, "Optically active 1,3-dioxin-4-ones" (EP 0500364 family). Different jurisdiction, different subject, no US authority prefix.
• Florida LLC FEI No. 93‑4728591 (High Seasons Ventures LLC) — a tax ID, not a patent.
• 0.4728591.000000 in US 2014/0092128 A1 — a color/CG parameter.

None of these are prior art to US 4,728,591 and I have excluded them.

---

1. Scope caveat — read this before the tables

"Patent citation" is ambiguous and the answer depends on which direction you mean:

• Backward citations (references cited in US 4,728,591) — the prior art considered against the claims. This is what § 102 analysis requires, so this is what I analyze below.
• Forward citations (later patents citing US 4,728,591) — these are not prior art (they post-date 1986‑03‑07). I list them separately in §5 to avoid contaminating the § 102 analysis.

Important limitation I must flag honestly: the front-page examiner "References Cited / U.S. Patent Documents" table did not render in any retrieved copy; the Google Patents page body I obtained contains the specification, abstract, claims metadata and classification data but not the citation table. The references below are the ones I could verify from the patent's own text, which Google Patents indexes under the "cited by/in the description" categories. They are genuine prior-art references, but my list may be incomplete — the examiner's table could contain additional items I have not seen. I did not fabricate entries to fill the gap.

---

2. U.S. patent documents cited in the specification

These are the only two U.S. patent citations I could confirm. They appear in the Background, col. ~"Description of the Prior Art," in the sentence: "While serial writing of such a pattern on a nanometer scale is possible, using known methods such as focused electron beams… See, for example, U.S. Pat. Nos. 4,103,064 and 4,103,073…"

Honest gap: I could not retrieve the titles or the exact issue dates of '064 and '073 from the searches run. I am reporting them literally as numbers, in the context in which the applicant placed them. Do not treat the "1978" inference as verified.

---

3. Non-patent literature cited in the specification

These are the substantive prior-art references and, in § 102 terms, they matter more than the two patents.

---

4. § 102 anticipation analysis (claim-by-claim)

Legal frame applied: under 35 U.S.C. § 102 a reference anticipates only if it discloses every limitation of the claim, arranged as in the claim. A reference that discloses only a sub-element is § 103 art (obviousness), not § 102 art. Sections 2–3 above are quoted by the applicant itself as the state of the art, which makes them admitted prior art.

Bottom line on § 102

No reference I could confirm discloses all limitations of any independent claim (1, 2, 3, 5, 6, 7). The applicant appears to have pre-empted the § 102 attack by drafting each independent claim to require both (a) a self-assembled 2‑D ordered array and (b) a substrate-interaction structure (patterned layer "determined by," electrode, depressions, or determining means). The confirmed prior art supplies (a) — N7/N8/N9/N10 — or a patterning concept — N1/N2/N3/N4/N12 — but not the combination.

The real exposure is § 103, and the dangerous combination is: Carter ('83/'84) or Tucker ('84) (use a biological molecule as a nanometer mask) + Keegstra & van Bruggen ('80) or Michel ('80) (self-assembled 2‑D protein arrays exist and can be bound to substrates) + Zingsheim ('77) (patterned surfaces direct protein adsorption). The patent's nonobviousness argument has to rest on the ordered, self-assembled, nondenatured character and the demonstration that the array can transfer its pattern to an overlayer (FIG. 4's actual metal-deposition result).

---

5. Forward citations — NOT prior art (listed to prevent confusion)

These later documents cite US 4,728,591 and therefore post-date it; they can never be § 102 art against it:

• US 4,802,951 A (Clark et al., 1989‑02‑07) — the applicant's own continuation ("Method for parallel fabrication of nanometer scale multi-device structures"), priority 1987‑12‑23. Sibling, not prior art.
• US 6,713,238; CA 2 328 593 A1; US 6,248,529; US 7,553,369; US 7,900,478; US 6,998,228; US 5,384,973 — all post-1988 forward citations.

Also note: WO 02/033461 A3 (Ozin et al.) cites US 4,728,591 at col. 5, lines 32–54 as an "X" reference against its claims 1–4 and 30–32 — a useful third-party read on which portions of '591 are considered the inventive core (the paragraph describing the array + overlayer + hole-transfer concept).

---

6. Uncertainty and what would close the gap

1. Incomplete citation list. I did not obtain the examiner's front-page "References Cited" table. The items in §2–§3 are verified from the patent's own text but may not be the full § 102 record. The authoritative list is on the face of the printed patent and in USPTO PatentCenter for Appl. No. 06/837,376.
2. US 4,103,064 / '073 titles and dates unverified. Do not cite dates or titles for these from this report.
3. Method claims (~23–38) not retrieved — consistent with the gap flagged in the earlier section of this analysis. My § 102 treatment of N11, N12, N13 and the "binding" methods is therefore partial.
4. Legal-status caveat carries over from the earlier section: "Expired – Fee Related," anticipated expiration 2006‑03‑07, is Google Patents' characterization and not independently verified — and it is irrelevant to § 102, which is measured against the 1986‑03‑07 filing date.
5. No non-patent public-use/on-sale analysis was attempted; if the inventors' own 1984–85 publications (e.g., the Clark/Douglas/Rothschild work) predate 1985‑03‑07, they would be applicant's-own § 102(b)/§ 103 art and could matter.

Sources: https://patents.google.com/patent/US4728591/en · https://patents.justia.com/patent/4728591 · https://uspto.report/patent/grant/4728591 · https://patents.google.com/patent/US4802951/en · https://patents.google.com/patent/US6713238/en (forward-citation table) · https://patentimages.storage.googleapis.com/a0/ef/8c/ff9017ceb05c6e/WO2002033461A3.pdf (third-party "X" citation of '591, col. 5:32–54) · http://data.epo.org/gpi/EP0500364A1 (to exclude the unrelated JP 4728591 A).

I'll ground this analysis by pulling the actual prior-art citations and related references for US 4,728,591.

Let me pull the specific prior-art references cited both in the patent's background and in its front-page citation list.

Obviousness Analysis — U.S. Patent No. 4,728,591 (35 U.S.C. § 103)

Scope, standard, and caveats first. This is a technical, retrospective analysis, not a legal opinion. The patent's priority date is 1986-03-07 (pre-AIA), so the governing provision is pre-AIA § 103(a), applied through the Graham v. John Deere, 383 U.S. 1 (1966), factors as refined by KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007). The patent is Expired – Fee Related (anticipated expiration 2006-03-07 per Google Patents), so the practical vehicle for this analysis would be an ex parte reexamination or a historical validity dispute, not a live infringement defense. I am building on the previously generated summary and litigation sections and do not repeat them.

Two grounding caveats that materially affect what follows:

1. The "Prior Art section" of the Google Patents page did not surface a discrete front-page References Cited list in what I retrieved. My tool run terminated at the step limit before I could pull the patent's cited-references table. Accordingly, my prior-art inventory below is drawn from (a) the patent's own "Description of the Prior Art" and in-specification citations, which are authoritative in the full text supplied, and (b) the working search results. Where I rely on my own knowledge of a reference I could not verify in this session, I say so explicitly.
2. The method claims (roughly 23–38) are still not available to me (flagged in the prior summary as a gap). I therefore analyze them only at the genus level via the disclosed process steps, and I mark any conclusion about them as provisional.

---

1. Person of ordinary skill in the art (PHOSITA)

For a March 1986 filing, the PHOSITA would be a researcher with a Ph.D. in physics, electrical engineering, materials science, or biophysics, or an M.S. with ~3–5 years' experience, working at the intersection of sub-micron lithography (e-beam writing, ion milling, resist processing) and structural biology / electron microscopy of two-dimensional protein crystals. Both skillsets existed in 1986 — the patent's own background section cites literature from both camps side by side, which is itself evidence that the two fields were being deliberately bridged before the filing.

---

2. Claim construction of the limitations that drive the § 103 fight

---

3. Prior-art inventory and what each reference teaches

Grouped by what it supplies to a combination. Citations marked ★ are quoted or referenced in the patent's own prior-art/background text supplied above; those are the most defensible to rely on.

A. Nanometer-scale patterning by serial writing (the problem the invention solves)

• ★ U.S. Pat. Nos. 4,103,064 and 4,103,073 — conventional e-beam/resist-based pattern definition (per the patent's background).
• ★ Craighead et al., "Ultra-Small Metal Particle Arrays Produced by High Resolution Electron-Beam Lithography," J. Appl. Phys. 53(11):7186–7188 (Nov. 1982) — teaches arrays of small metal particles made by e-beam lithography. This is directly material to claim 2's "plurality of projections," independent of how they are made.
• ★ Mochel et al., "Electron Beam Writing on a 20-Å Scale in Metal β-Aluminas," Appl. Phys. Lett. 42(4):392–394 (1983) and Isaacson, "Electrons, Ions and Photons in Submicron Research" (1984) — teach that 1–10 nm feature definition, and ion-beam pattern transfer, were known. Isaacson in particular supplies the ion-milling/etching transfer step used in the claims.

B. The decisive reference — biological molecules as nanometer lithographic masks

• ★ Carter, "Molecular Level Fabrication Techniques and Molecular Electron Devices," J. Vac. Sci. Technol. B1(4):959–968 (Oct.–Dec. 1983) — I verified this reference in-session (Semantic Scholar record). Its abstract explicitly addresses "techniques of building up from the molecular level," nanometer-scale lithography, and states that "three approaches possible involving biological and Langmuir-Blodgett materials will be described." The patent itself characterizes Carter as suggesting "large biological molecules may be used as a mask to apply a nanometer scale pattern of material on a substrate … includ[ing] DNA and denatured proteins." Carter also discusses electron tunneling in short periodic arrays — which is the physics behind the SIS-junction and quantum-interference embodiments.
• ★ Carter, "Biotechnical Synergism in Molecular Electronics," Nonlinear Electrodynamics in Biological Systems, Plenum Press, pp. 260–273, and Tucker, "Biochips: Can Molecules Compute?," High Technology 4(2):36–47, 79 (Feb. 1984) (esp. p. 46) — popularize the same "protein as mask / molecular device" concept and, critically, evidence that this idea was in the air before the 1986 filing.

C. How to make the ordered array

• ★ Keegstra & van Bruggen, "Electron Microscopy at Molecular Dimensions," Springer-Verlag, New York (1980), pp. 318–327 — the patent's own cited "general method for production of such a self-assembled protein array," including arrays of Panulirus interruptus hemocyanin that the patent describes as having "an ordered pattern of through holes." This is the single most damaging reference: it supplies the "two-dimensionally ordered molecular array with a 1–50 nm pattern of holes" element.
• ★ Michel et al., "Electron Microscopy at Molecular Dimensions," Springer-Verlag (1980), p. 27 — isolation of the S-layer of Sulfolobus acidocaldarius, a 10 nm-thick monolayer with a 22 nm lattice and ~70% void volume — i.e., a periodic stencil.
• ★ Scott et al., "Two-Dimensional Crystals in Detergent-Extracted Disk Membranes From Frog Rod Outer Segments (ROS)," Bio. Phys. J. 33:293a (1981) — supplies rhodopsin 2-D crystals (claim 10).
• ★ Wright et al., "Localizing Active Sites in Zeolitic Catalysts," Nature 318:611 (1985) — supplies non-biological periodic arrays of nanometer-dimension holes (zeolites), which undercuts any argument that the "self-assembled ordered array" genus is limited to biology.
• Deckman & Dunsmuir, "Natural lithography," Appl. Phys. Lett. 41(4):377 (1982) — my own knowledge; I could not verify this in-session because the tool run terminated. It teaches using a self-assembled ordered array of colloidal spheres as a lithographic mask for pattern transfer. If verified, it is a clean § 103 reference against the broad genus claims and it removes the "biological mask" novelty entirely. Treat this as unverified.

D. Selective/ordered binding of the array (claims re electrostatic binding, patterned activation)

• ★ Zingsheim, Scanning Electron Microscopy, Vol. I, pp. 357–364 (1977) — teaches inhomogeneous protein adsorption on a patterned carbon film, with "3–4 nm side contamination lines at spacings of about 50 nm." This supplies the motivation for every "bind the array selectively to activated micropatterned areas" limitation.

E. Readout / state-determination (claims 3–4, and the bR memory embodiment)

• ★ Keszthelyi, "Orientation of Purple Membrane by Electric Fields," Methods in Enzymology, Vol. 88, pp. 287–297 — electric-field ordering and electrical interrogation of purple membrane.
• ★ Binnig, Rohrer et al., "7×7 Reconstruction on Si(111) Resolved in Real Space," Phys. Rev. Lett. 50:120 (1983) — the STM, which the specification itself invokes as the addressing/readout mechanism for the memory embodiment.

---

4. The § 103 case, by claim group

Claims 1, 6, 7 (substrate + ordered array + pattern-determined overlayer [+ electrode])

Combination: Carter 1983 + Keegstra & van Bruggen 1980 + Isaacson 1984 (or Craighead 1982).

• Carter supplies the motivation and the genus: use a large biological molecule as a mask to impose a nanometer pattern on a substrate.
• Keegstra supplies the species that makes Carter's idea actually periodic: a self-assembled 2-D protein crystal with an ordered array of through-holes having ~1–50 nm character.
• Isaacson/Craighead supply the deposition + ion-milling transfer step, and Craighead supplies the resulting array of metal islands.

Articulated motivation (KSR "rational underpinning"): Both references are in the same field (nanometer-scale pattern formation). Carter expressly identifies the reason a biological mask is attractive — nanometer-scale features are otherwise only obtainable by slow serial writing. Keegstra's crystals are the most regular nanometer-scale templates then known, and the patent itself concedes the whole point: "A particular advantage of employing nondenatured protein molecules is their extremely regular nature." A PHOSITA seeking a periodic array of metal islands (Craighead's goal) would predictably reach for the most periodic available mask. Claim 1's "electrode" limitation is met by conventional contact metallization, which the patent itself concedes is "fabricated … using conventional methods." Under KSR, "the combination of familiar elements according to known methods is likely to be obvious when it does no more than yield predictable results."

Where the patent pushes back (and the counter is decent): Carter's masks are "denatured proteins" and DNA. A denatured protein is a flattened, unfolded sheet that shadows; a nondenatured crystal is a 3-D folded, porous structure that acts as a stencil. One can argue Carter teaches away from nondenatured material — if you need the protein to lie flat as a mask, you deliberately destroy its ordered native structure. This is the strongest nonobviousness argument for claims 1/6/7/8. It is weakened by the fact that Keegstra and Michel, both pre-1986, make the through-hole stencil obviousness plainly available as an alternative mode.

Claims 2 (projections + dissimilar material between them, "not naturally formed")

Combination: Craighead 1982 (arrays of metal particles) + conventional IC planarization/insulator deposition.

The claim adds only a fill step (the specification's FIG. 7D "second dissimilar material 34 such as an insulator"). Isolating metal islands with a dielectric is standard semiconductor practice and is exactly what the parallelism rationale requires (isolated islands → SIS junctions, superlattices). KSR's "predictable variation" and "obvious to try" rationales apply cleanly. Likely obvious.

Claims 3–4 (means for determining the state of selected portions of the array; electrodes)

Combination: Keszthelyi + Binnig/Rohrer 1983 + the known bR photoelectrical response.

The specification itself concedes that bR's voltage-dependent 570→610 nm shift "has been demonstrated" (i.e., it is prior art, not the inventors' discovery) and that STM addressing "has been developed by Binnig, Rohrer et al." Claim 3 is a means-plus-function claim (pre-AIA § 112 ¶ 6), so its scope is limited to the disclosed structures (electrodes, STM probe, laser) and equivalents — every one of which is in the cited prior art. Likely obvious, though a zealous patentee would argue the combination of a periodic bR array with per-site addressing was not suggested; that is a genuine but modest argument.

Claim 5 (substrate base with depressions corresponding to the array's holes)

Combination: Keegstra/Michel (template) + Isaacson (ion milling/etching). Pitting a substrate through a mask is the definition of the transfer step the art already used. Plainly obvious.

Claims 8–14, 17 (nondenatured proteins; S-layer; purple membrane; rhodopsin; cytochrome oxidase; porin; gap junctions; holes 1–50 nm)

These are species selections from a known genus. Once claim 1's genus is obvious, the KSR principle that "when a patent claims a structure already known in the prior art … the burden of showing nonobviousness is high" and the "obvious to try" line of reasoning apply: the patent itself states "Approximately 1000 distinct two-dimensional crystallizable proteins are known," and it cites the specific isolation protocols for the S-layer (Michel) and rhodopsin (Scott). Choosing among ~1000 known 2-D-crystallizable proteins, each with a known lattice constant in the 1–50 nm range, is routine selection with a reasonable expectation of success. Likely obvious.

Claims 15–16 (surface layer of a 3-D crystal; protein crystal)

The specification cites lysozyme and hemoglobin 3-D crystals as known. Using a crystal surface's contour as a template is a new-use/application step that the prior art does not squarely suggest (the cited art uses monolayer 2-D crystals). Moderate nonobviousness risk for an examiner — this is the kind of claim that survives on the strength of "the prior art masks are flat; this one uses topography."

Claim 18 (chemical-reactivity-variation pattern — the bR local-pH-gradient mechanism)

This is the patent's most defensible subject matter. The claimed patterning agency — using the spatially localized, light-driven proton pumping of bacteriorhodopsin to generate a local pH gradient at each molecule and thereby activate pH-sensitive reactions in an overlayer at nanometer scale — is not, on the record I retrieved, suggested by any single reference or combination. It requires: (i) knowledge of bR proton pumping (known), (ii) knowledge of pH-sensitive reactions (the patent concedes "a large number of reactions are known which are pH sensitive"), and (iii) the insight to use (i) to spatially pattern (ii) at molecular resolution. That third step is where a § 103 attack is weakest — it is a new principle of operation, not a rearrangement of known parts. KSR cautions against "hindsight" reconstructions, and this is precisely the fact pattern where hindsight is most dangerous. Probably nonobvious. (It remains attackable as "obvious to try" if a reference had ever proposed chemical-reactivity-based nanolithography, but I found none.)

Claims 19–21 (multilayer substrates; commensurate/incommensurate patterns)

Multilayer/MBE substrates are conceded as known ("Layers of substrate 10a may be single crystals of mixed atomic layers as formed by molecular beam epitaxy"), and commensurate vs. incommensurate registry is a standard epitaxy concept. The novelty lies in applying it at nanometer scale with a protein template — a "new use of a known technique" that KSR disfavors as a basis for patentability but that is not automatically obvious either. Mixed; lean obvious on the strength of KSR's "improve similar devices in the same way" rationale.

Claims 23–38 (method claims — text not retrieved)

Provisional read: the disclosed method is (i) bind a self-assembled 2-D array to a substrate; (ii) deposit/overcoat or use the array's holes as a deposition or etch mask; (iii) transfer the pattern by ion milling/etching/X-ray/e-beam; (iv) strip the biological array (detergent/protease/acid). Steps (ii)–(iv) are standard lithography (Isaacson; the patent's own background), and step (i) is Keegstra/Michel plus the binding techniques the patent concedes are conventional. Provisionally obvious via Carter + Keegstra + Isaacson; but I cannot finalize this without the actual claim language.

---

5. Motivation to combine — the consolidated argument

A single, coherent KSR-compliant motivation runs through all of the above:

1. Same field of endeavor. Carter (biological masks for nanolithography) and Keegstra/Michel (2-D protein crystal production) are both directed to nanometer-scale ordering; Craighead/Isaacson supply the transfer chemistry of the very same art.
2. The problem was known and articulated in the art. Serial e-beam writing of "over 10,000 devices in a micron square area" was recognized as impractically slow — the patent's background says so, and Carter's stated goal is to avoid it. That is a long-felt need, which normally cuts for obviousness (it supplies the "reason to combine").
3. Reasonable expectation of success. The lattice parameters (22 nm S-layer; colloid arrays 5–1000 nm; zeolite pores) were known and within the claimed 1–50 nm range; ion-milling through a mask to punch holes was known; lift-off and selective adsorption were known (Zingsheim).
4. Only a finite, identified set of candidate masks existed (2-D protein crystals, zeolites, colloidal crystals) with predictable properties — the hallmark of "obvious to try."

---

6. Why the patent is nonetheless not trivially obvious

Counterweights a challenger must overcome:

• Teaching away / different mode of operation. Carter's exemplars are denatured proteins and DNA — non-periodic shadows. The invention's core insight is that a nondenatured, self-assembled crystal — a periodic stencil with through-holes — is a better mask. Arguably the art steered away from ordered biologics by degrading them first.
• No pre-1986 reference actually reduces the invention to practice. I found no reference that used a biological 2-D crystal as a lithographic stencil before the filing. Keegstra characterized crystals; Carter proposed molecules-as-masks; the combination that makes the periodic stencil work appears to be the inventors' contribution.
• Claim 18 (chemical-reactivity patterns) is a genuinely new principle of operation, and it is the claim most resistant to § 103.
• Secondary considerations are essentially absent from the record. There is no evidence of commercial success, licensing, copying, or unexpected results that I could find — and the patent's own litigation summary shows it was never asserted and expired for fee nonpayment. That absence of secondary indicia weakens the patentee's side of the Graham balance, not strengthens it.

---

7. Bottom line, claim by claim

Net assessment: A well-constructed § 103 challenge would likely invalidate the apparatus and method claims directed to the basic substrate/array/overlayer/electrode architecture (1, 2, 5, 6, 7, 8–14, 17, 19) on Carter 1983 in view of Keegstra & van Bruggen 1980 and Isaacson 1984, with Craighead 1982 for the island-array claim and Wright 1985 foreclosing any non-biological breadth argument. The claims that would most plausibly survive are claim 18 (chemical-reactivity-variation patterning) and, secondarily, claims 15–16 (3-D crystal surface as template). Whether the "nondenatured" limitation in claim 8 rescues the genus claims depends on how strongly one reads Carter's use of "denatured proteins" as teaching away — a close call.

What I could not do, and what would change this analysis: (i) I never retrieved the patent's front-page References Cited list, so there may be cited U.S. patents (e.g., ordered-array or resist patents) that make the case stronger or weaker; (ii) I never retrieved the method claim text (23–38), and their scope could be broader or narrower than assumed; (iii) Deckman & Dunsmuir's "natural lithography" (1982) is from my own knowledge and is unverified here — if genuine and pre-1986, it is the cleanest single-reference attack on the broad "self-assembled ordered array as lithographic mask" concept and would meaningfully strengthen the § 103 case. Verifying it in EPO/Google Patents is the highest-value next step.

Sources consulted in-session: US4728591 on Google Patents (authoritative full text, including the "Description of the Prior Art"); Carter 1983 record; US4802951 (continuation). All other references are as cited within the patent text supplied, plus the explicitly flagged unverified items.