US 5583727

I'll search for authoritative information on US 5583727, including any litigation activity.

Let me check specifically for litigation activity and any CAFC dockets.

US 5583727 — Summary

Important preliminary note on search scope: I searched for litigation and CAFC 2026 docket activity. I found no evidence that US 5583727 has ever been litigated, and no CAFC 2026 docket involving it. The only "5583727" hits in USPTO-hosted documents were examiner search-strategy logs (patentability searches), not dockets. I cannot confirm a negative with certainty, but there is no positive indication of any court proceeding. Treat "no litigation found" as a reported absence of evidence, not a verified fact.

Bibliographic data (from the authoritative Google Patents full text)

Abstract

A magnetic recording data storage system uses a medium substrate with multiple data layers and an MR read sensor providing a direct digital output while reading all data layers simultaneously. In a disk drive embodiment, two magnetically isolated/decoupled data layers are written independently by selecting coercivity of the magnetic materials and write current strength. In one embodiment, the layers have different coercivity-vs-temperature dependencies and the disk is heated (or not) before writing with a field affecting only one layer. The magnetic field strength from written data in each layer differs so the data can be distinguished. The preferred sensor is a GMR multilayer of alternating ferromagnetic and nonferromagnetic metal spacer layers, in which each ferromagnetic layer's magnetic moment responds to a different external field strength, allowing independent switching of magnetization direction.

Independent claims — plain language

The patent has 28 claims; claims 1 and 16 are independent. (Note: claim 1 as published contains a typographical error — "aid magnetic flux superposition" for "said"; I have not corrected it and read it literally as "said.")

Claim 1 — The storage system (method-agnostic apparatus claim):
A magnetic data storage system with four cooperating parts:

1. Media — a substrate with first and second magnetic data layers, plus a magnetic decoupling layer between them; the first layer's material has a coercivity different from the second's.
2. Write head — positioned close to the media during relative motion; generates a write field in the second layer that is less than the coercivity of the first layer, so data can be written to the second layer without altering data in the isolated first layer.
3. MR read head — positioned close during relative motion; senses fields from both layers simultaneously; exhibits discrete steps of decreasing resistance in response to discrete steps of increasing magnetic field from the superposition of flux from both decoupled layers (the superposition representing the data in both layers).
4. Sensing circuitry converting resistance to an output signal, plus logic circuitry decoding that signal into the separate digital data written in each layer.

Claim 16 — The disk drive (more structurally specific species of claim 1):
A magnetic recording disk drive comprising:

• a disk with substrate and first/second magnetically isolated, independently recordable data layers, the first layer's coercivity differing from the second's;
• a motor rotating the disk;
• an inductive write head generating a write field in the second layer less than the first layer's coercivity (so the first layer is not altered);
• a GMR MR read sensor comprising a substrate with an essentially single-crystal face, a seed layer, and a multilayer stack of alternating permalloy and copper spacer layers (at least three permalloy layers, two Cu spacers), neighboring permalloy layers antiferromagnetically coupled across intermediate Cu, each with antiparallel moment alignment at zero external field, and each permalloy layer having a substantially uniaxial magnetic anisotropy different from the others, so each switches at a different field — producing the discrete resistance steps used to represent the superposition of flux from both layers;
• sensing circuitry and logic circuitry as in claim 1;
• a carrier supporting the sensor, an actuator moving the carrier radially across the disk, and means for supporting the motor and actuator.

Dependent claim coverage (brief)

• Claims 2–4, 18, 20–21: first layer contains pre-written servo data (buried servo); in-plane (longitudinal) or perpendicular domain orientation.
• Claims 5–15: read-head details — ≥3 antiferromagnetically coupled ferromagnetic layers with uniaxial anisotropy and layer-specific switching fields; 110 crystalline substrate face; MgO substrate; Fe/Pt seed bilayer; permalloy layers; Cu spacers; differing anisotropies (12, 25); differing interlayer coupling field strengths (13, 26); differing spacer thicknesses (14, 27); bias layer fixing the nearest/farthest layer's magnetization (15, 28).
• Claims 17, 19: heater for heating data layers before writing; magnetic decoupling layer between the two data layers.

Technical significance

This is an early IBM/Parkin disclosure combining (a) multi-layer, magnetically decoupled recording media with (b) a digital-output GMR read sensor whose ferromagnetic layers switch at distinct fields (via differing strain-induced uniaxial anisotropy). It also discloses thermally assisted writing via differing coercivity-vs-temperature dependence and a heat source, a concept later cited in the thermal-assist literature (e.g., Suess US 11,133,031 lists "5583727 | December 10, 1996 | Parkin"). The 1999 Navy confirmatory license indicates government-funded work.

Uncertainty flags

• No litigation/CAFC 2026 activity found for 5583727 — this is an absence of evidence from my searches; I could not query every docket system directly.
• Bibliographic data, abstract, and claims are taken from the authoritative Google Patents full text of US5583727A; the patent's own text repeatedly notes legal-status and priority-date fields are assumptions by the source, not legal conclusions.
• Current assignee listed as IBM; the Navy entry is a confirmatory license, not an ownership transfer — do not treat the Navy as assignee.

I'll search litigation databases and other sources for US patent 5583727.

Bottom line

I found no litigation (or PTAB/post-grant proceeding) involving U.S. Patent 5,583,727. I ran multiple targeted searches for the exact number and for the patent in a litigation context. Every hit for "5583727" was a citation, prior-art reference, or prior-art search-string entry — not a case in which the patent was asserted or challenged.

What I searched and what came back

Requested table

Context that bears on why no litigation should be expected

The patent's own record (authoritative full text provided) shows:

• Filing date / grant: App. No. 08/441,100, filed 1995‑05‑15, granted 1996‑12‑10.
• Legal status: Expired – Fee Related (Google Patents) with an anticipated expiration of 2015‑05‑15. A 20‑year utility term means any infringement suit would have had to be filed on or before mid‑2015; the window is closed.
• Assignee: International Business Machines Corp. (IBM). Under IBM's general practice, patents of this subject matter were licensing tools, not litigation vehicles.
• Government interest: On 1999‑01‑07 a confirmatory license was recorded to the Secretary of the Navy, United States of America (assignor IBM Corporation). This is a statutory confirmatory instrument acknowledging Government rights/funding — it is not litigation and should not be confused with one.
• Family/related filing: The specification states this application and concurrently filed Ser. No. 08/441,133 ("Digital Magnetoresistive Sensor Based on the Giant Magnetoresistance Effect") share a common specification. If any enforcement activity existed, the sibling application is the likelier candidate — worth checking separately (likely issued as an IBM patent in the same family; I did not confirm its number).
• Third-party citations: The patent is cited by 54 later publications (per the Google Patents "Cited By" list), including HDD-related work by Seagate, Hitachi Global Storage Technologies, and others — citation, not assertion.

Caveats and limits of this search

• I was working from web search, not a direct live query of PACER, the Unified Patents litigation portal's case list, or RPX Insight's per-patent page. I attempted the RPX/insight and court-complaint queries, but my tool budget was exhausted before those returned.
• Web search is weak at surfacing old (pre‑2005) district court dockets in which a patent number may appear only in an exhibit or claim chart rather than in a searchable caption.
• The absence of results is not proof of absence of a suit. It is a statement that no public record of litigation came back in these searches.

Recommended verification steps (to close the gap)

1. Unified Patents litigation portal — search the patent number directly: https://portal.unifiedpatents.com/litigation/patent/5583727 (or the caselist search).
2. RPX Insight — per-patent page: https://insight.rpxcorp.com/patents/US5583727 ("Litigation" tab).
3. CourtListener RECAP — full-text search: https://www.courtlistener.com/?q=%225583727%22 and %225%2C583%2C727%22 (both formatting variants).
4. USPTO Patent Center / Patent Trial and Appeal Board (PTAB) and Patent Public Search for any ex parte reexamination, IPR/PGR, or reissue — note that any such proceeding would have had to occur before the 2015 expiration.
5. PACER docket search across districts (IBM-home districts: S.D.N.Y., N.D. Cal., D. Del., E.D. Tex.).

If those confirm no hits, the correct representation to a court or client is: "No litigation involving U.S. Patent 5,583,727 has been identified; the patent expired on May 15, 2015."

I'll verify the structured data against public sources before writing anything up.

Let me check a few more targeted angles, including ex parte reexamination (a non-AIA, non-PTAB route that sometimes surfaces for older patents).

Proceedings overview

Total AIA trial proceedings on file: 0. The USPTO Open Data Portal structured block returns no AIA trial proceedings for US 5,583,727, and my independent web searches surfaced none — no IPRs, no PGRs, no CBMs, no institution decisions, no Final Written Decisions, and no Federal Circuit appeals arising from a PTAB trial on this patent. The breakdown is therefore 0 active / 0 invalidated / 0 sustained / 0 settled / 0 denied-by-institution. My searches for the patent number returned only citations to the patent as prior art inside unrelated proceedings' search histories (e.g., the USPTO's own examiner search-strategy logs showing ("5583727") queries in IPR petitions on other patents) — a citation hit is not a proceeding on this patent, and I have not attributed any to it.

Defensive posture: This is not "the patent survived two IPRs and is hardened," and it is not "claims 1–5 were canceled." It is a third-party-never-challenged posture — and, more decisively, the patent is expired. Because the patent expired on its 20-year date (2015-05-15, per the anticipated-expiration entry in the structured data; status: "Expired - Fee Related"), a defensive IPR today is largely moot: the Board will not institute where the patent has expired and the only live remedy would be past damages, and there is no patent owner to negotiate with for anything other than historical royalties/back-damages. The real defensive value here is the calendar, not the PTAB record. See "Recommended next steps."

Proceedings

None to report. There is no proceeding number to list, and I will not invent one. For completeness, the four AIA-trial items the template asks about are answered as follows:

• Type: N/A — no petition of any AIA type was ever filed.
• Filed: N/A.
• Status: N/A. (The patent's own legal status per the structured data is "Expired - Fee Related," with anticipated expiration 2015-05-15.)
• Judge panel: N/A.
• Petition grounds: N/A.
• Institution decision: N/A.
• Final Written Decision: N/A.
• Settlement / termination: N/A.
• Appeal: N/A — no FWD to appeal, and I found no Federal Circuit docket (CourtListener) involving this patent.
• Defensive value: None derived from PTAB activity; see strategic summary.

Why the absence is expected — and partly structural

The patent's effective filing date is 1995-05-15, well before the AIA's 2013-03-16 threshold. That forecloses two of the three AIA trial types by statute:

• Post-Grant Review is available only for patents with an effective filing date on or after 2013-03-16, so PGR was never a possibility.
• Covered Business Method review was, by definition, limited to patents claiming a financial-services method; this is a magnetic-recording/GMR-read-sensor patent, and CBM review also sunset on 2020-09-16.
• IPR was the only AIA avenue ever available, and it never materialized.

That "never materialized" is a meaningful signal in context: this is a landmark IBM/Parkin GMR patent that has been cited by a large set of later magnetic-media and heat-assisted-magnetic-recording patents (the "Cited By" list in the structured data runs to 54 entries, including Hitachi GST, Seagate, HGST, Samsung, Fujitsu, and Western Digital-related work). Being heavily cited while never being IPR'd is more consistent with a patent that was never commercially asserted in a way that justified the cost of a petition than with one that was tested and survived.

Strategic summary

Claim status: all 28 claims UNTESTED. Claims 1–28 have never been construed or adjudicated by the PTAB. There are no canceled claims, no claims held unpatentable, and no claims confirmed in a Final Written Decision. Any representation to the contrary — e.g., a demand letter asserting that the claims were "upheld" or "invalidated" at the PTAB — is wrong on the record. Independent claims 1 and 16 (the storage system and the disk drive) are the core assertions to analyze on the merits; the dependent claims worth reading closely are 5–15 (GMR read-head limitations, including the 110 MgO substrate of claim 6/7, the Fe/Pt seed layer of claim 8/9, and the per-layer anisotropy/coupling distinctions of claims 12–15) and 17–28 (the disk-drive analogues).

Estoppel landscape: clean. Because no IPR ever reached a Final Written Decision, § 315(e)(2) estoppel never attached to anyone. There is no petitioner, no privy, no real party in interest who is barred from raising anything. A defendant today has the entire prior-art universe available: printed publications and patents under §§ 102/103, system prior art, and — critically for this patent class — the § 112 written-description/enablement attack on the broad functional claim language ("discrete steps of decreasing electrical resistance… logic circuitry for decoding"). Equally, no one else's failed IPR gives you a roadmap or a place to piggyback; you would be building the invalidity case from scratch, which is a real cost.

Pattern signals: none. No petitioner filed multiple IPRs against this patent (there is no petitioner at all). The patent owner never had a PTAB appeal to pursue aggressively. There is no defensive aggregator (Unified Patents or similar) in the chain — the only third-party interest recorded is a 1999-01-07 confirmatory license to the U.S. Navy, which reflects government funding/rights rather than a validity challenge. The only "challenge-like" events in the file history are routine assignments: IBM as original assignee (1995-07-05) and the Navy confirmatory license (1999-01-07).

Two caveats I want to be explicit about, per the don't-fabricate rule. (1) I searched for ex parte reexamination activity and found none, but reexamination records are not surfaced cleanly by general web search and I could not exhaustively confirm a negative; if a reexam certificate exists it would not be a PTAB proceeding anyway and would not change the analysis below. (2) The structured ODP block is stated to reflect the most recent ingest; a very recently filed petition would not yet appear there and I found no web evidence of one — but the practical significance of any such petition is minimal given expiration.

Recommended next steps

1. Lead with the expiration, not the PTAB record. The patent's 20-year term ran from the 1995-05-15 filing date, and the structured data shows anticipated expiration on 2015-05-15 with status "Expired - Fee Related." There is no PTAB activity to point to, but you may not need it: confirm the expiration and any post-2015-05-15 chain of title before engaging on the merits, and check whether any assertion you face is for pre-expiration conduct only. An assertion resting on post-expiration sales is a non-starter, and a demand letter that cites claims 1–28 without acknowledging expiration is a strong tell that the correspondence is not carefully prepared.
2. If assertion is for historical conduct, do not assume you need to file an IPR — you almost certainly do not. On an expired patent the Board's practice is to decline to institute where there is no live controversy to resolve, and the practical fight is damages, not validity. Preserve § 282 invalidity defenses in the litigation and consider a stay/expiration-based motion practice.
3. If for some reason you do want a Board ruling, note the statutory milestones that would govern a hypothetical IPR: institution decision due within 6 months of the petition's filing (35 U.S.C. § 314(b)), trial completing with an FWD within 12 months of institution (§ 316(a)(11)), extendable up to 6 months for good cause by the Director. There are no pending proceedings, so there are no live institution/oral-hearing/FWD dates to track.
4. Do not cite a non-existent FWD. Because there is no FWD, there is no document to link to on PTAB E2E. The best public anchors for the patent's own record are the Google Patents page (https://patents.google.com/patent/US5583727/en) and the USPTO PatentCenter/Global Dossier entries for US 08/441,100. The only "PTAB" URLs my searches returned for this patent number are unrelated petition documents whose examiner search logs happen to string-search the number (e.g., https://ptacts.uspto.gov/ptacts/public-informations/petitions/[1558577](/patent/1558577)/download-documents and .../1558578/...) — cite those only as evidence of third-party citation, never as proceedings on this patent.

Bottom line: no PTAB proceedings exist on US 5,583,727, so no claim has been canceled or sustained; § 315(e)(2) estoppel is entirely absent; and the patent's expiration in 2015 is the dominant defensive fact. Treat any representation of PTAB history for this patent as unsupported.

I'll attempt to pull the USPTO assignment records and cross-reference sources before writing.

Preliminary note on source access

I was not able to query the USPTO Assignment Center directly, and no indexed page surfaced the underlying assignment abstract for this patent. What I can verify is the set of assignment-derived legal events carried by Google Patents (which are themselves sourced from the USPTO assignment database) plus the bibliographic records in the authoritative full text. Reel/frame numbers and the correspondent of record are not exposed in any source I reached, and I will not invent them. Where the required fields are unavailable I say so explicitly rather than guess — this matters for signal #3 ("repeat correspondent"), which is therefore unclear by construction, not "not present."

---

Inventors

Basis: the patent names only Parkin, and the recorded assignment of assignor's interest from "PARKIN, STUART S. P." to International Business Machines Corporation (recorded 1995-07-05, six weeks after the 1995-05-15 filing) establishes that he was under an obligation to assign to IBM at the time of filing. The specification's experimental basis — 110 MgO substrates, magnetron sputtering of Co/Cu and permalloy/Cu multilayers, an in-house citation to Harp and Parkin, Appl. Phys. Lett. 65(24), 3063 (12 Dec. 1994) — is consistent with IBM's GMR research group. I do not have a source in hand that states his specific IBM laboratory on the filing date, so I flag that as reported-but-unverified detail rather than assert it.

Unusual-pattern check: there is no multi-inventor dynamic here to check (single inventor), and no evidence of inventor departure timed to the filing. Parkin is publicly reported to have remained at IBM for nearly two more decades after this filing and to have left IBM around 2014 for a directorship at a Max Planck institute — that characterization is from general knowledge, not from a source I retrieved in this session, so treat it as low-confidence context. No fire-sale-precursor pattern.

Sibling application: the patent's own RELATED APPLICATIONS section discloses concurrently filed Ser. No. 08/441,133, "Digital magnetoresistive sensor based on the giant magnetoresistance effect," sharing this common specification. That application is part of the same ownership record and should be checked separately in the Assignment Center; I did not verify its issue number and will not state one.

---

Original assignee

International Business Machines Corporation (IBM) — named as the applicant at filing (1995-05-15) and as assignee on the recorded assignment of assignor's interest (recorded 1995-07-05).

• Primary line of business: diversified computer hardware, software and services; at the relevant time IBM was also a vertically integrated manufacturer of magnetic recording disk drives and, critically, of the GMR read heads that this patent's sensor disclosure feeds into.
• Did IBM ship a product embodying the claims? No product is known that embodies claims 1 or 16. The claims require a medium with two magnetically decoupled data layers read simultaneously by a multi-layer digital-output GMR sensor whose individual ferromagnetic layers switch at distinct fields. That is a research-stage system architecture. IBM did commercialize GMR read heads in the late 1990s, but a GMR head in a conventional single-data-layer drive does not practice these claims. The patent reads as a research disclosure, not a product specification; note also the experimental layers described (15 Å permalloy / 19.5 Å Cu on 110 MgO with a 5 Å Fe / 50 Å Pt seed) are epitaxial-laboratory structures, not manufacturable head stacks.
• Current status of the original assignee: operating. IBM continues as an operating company. Its disk-drive business was divested in the early 2000s (to Hitachi, forming Hitachi Global Storage Technologies, later folded into Western Digital), but the assignment records show no transfer of this patent out of IBM, so the divestiture did not carry this asset.

Per the authoritative full text, the current assignee is listed as International Business Machines Corp. The Navy entry below is a license, not an ownership transfer — do not record the U.S. Navy as assignee.

---

Assignment timeline

Only two assignment-database events are reflected in the records I could reach. Reel/frame numbers were not retrievable for either entry, and execution dates are not published in the indexed text — the Google Patents entry for the 1995 event reads only "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)," i.e. the document is referenced but not transcribed. I am reporting the dates that are published (recording dates) and marking everything else unknown.

• Execution date unknown / recorded 1995-07-05 — Reel not retrievable from available sources

  • Conveyance: Assignment of assignor's interest
  • Assignor: Parkin, Stuart S. P.
  • Assignee: International Business Machines Corporation
  • Correspondent: Not exposed in any source reached — cannot be read from the indexed record ("SEE DOCUMENT FOR DETAILS"). No recurrence call is possible; this is a data gap, not a negative finding.
  • Context: Routine pre-issue employee-to-employer assignment — the standard IBM inventor assignment, executed within ~7 weeks of filing and ~17 months before issue.

• Execution date unknown / recorded 1999-01-07 — Reel not retrievable from available sources

  • Conveyance: Confirmatory License
  • Assignor: IBM Corporation
  • Assignee: Secretary of the Navy, United States of America
  • Correspondent: Not exposed in any source reached. Again a data gap.
  • Context: Not an ownership transfer — a confirmatory license perfecting a pre-existing government license, consistent with federally funded research contracting (a nonexclusive, irrevocable government license rather than a conveyance of title). It is the classic tell that IBM's work on this subject matter was government-sponsored, and it is a genuine asset encumbrance: the U.S. Government holds license rights in this patent.

• 1996-12-10 — Issue/publication date (not an assignment event).

• 2015-05-15 — Anticipated expiration; legal status "Expired – Fee Related."

Note on term and status (flagging an internal inconsistency in the source data): the application was filed 1995-05-15, i.e. before the June 8, 1995 URAA cutoff, so the term is the longer of 17 years from issue (2013-12-10) or 20 years from filing (2015-05-15). The "anticipated expiration 2015-05-15" entry therefore reflects a full-term run, yet the status label "Expired – Fee Related" ordinarily denotes lapse for non-payment of a maintenance fee. These two entries are not comfortably reconcilable on the face of the record. I cannot resolve which is correct without the maintenance-fee event history, which I could not retrieve. Either way, the patent is expired and has been since 2015, so it is not assertable today.

No other recorded assignments exist in any source I reached — no transfer to an IP holding company, no security agreement, no merger conveyance, no change of name. Between 1999 and expiry there is a 16-year blank in the record, which is itself the finding: IBM appears to have held title from filing through expiration.

---

Timeline diagram

timeline
    title Ownership of US 5583727
    1995 : Application filed by IBM
         : Parkin assigns rights to IBM
    1996 : Patent issues as US 5583727
    1999 : Confirmatory license to US Navy
    2015 : Patent expires

---

NPE / troll-pattern signals

1. Shell-entity transfer — not present. No assignment to any entity with an "IP / Holdings / Licensing / Ventures" suffix, and no post-1999 assignment at all. The only non-IBM assignee on record is a U.S. government agency receiving a license, not title (recorded 1999-01-07).

2. Known asserter in the chain — not present. Neither recorded link involves Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Erich Spangenberg entities, or any other entity surfaced as a high-frequency plaintiff. No such name appears in the chain I could observe.

3. Repeat correspondent across the chain — unclear (data gap). Reel/frame documents are not retrievable through the sources available to me, and the 1995 entry is carried as "SEE DOCUMENT FOR DETAILS," so no correspondent name is exposed at all — not for the 1995 IBM assignment, not for the 1999 Navy confirmatory license. With zero correspondent data points I cannot assess recurrence. This signal is unresolved, not cleared. If you can pull the two PDFs from the Assignment Center, the correspondent fields on those filings are the single highest-value remaining item.

4. Cascading transfers — not present. Two recordings 3 years 6 months apart, both before 2000. There is no chain of LLCs and no sub-24-month sequence.

5. Pre-litigation transfer — not present. No litigation naming this patent was found (consistent with the prior section of this analysis, which likewise found no litigation or CAFC 2026 docket). With no suit on record there is no transfer to be measured against a filing date.

6. Bankruptcy fire-sale — not present. IBM has not been a Chapter 7/11 debtor, and no sale of this patent appears in any record. No Kodak/Nortel/Polaroid-style event touches this chain.

7. Privateering — not present. No NPE sits in the chain, so there is no operating company → NPE transfer capable of being a privateering arrangement.

8. Defensive aggregator — not present. The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN. Expiration, not neutralization-by-purchase, ended this asset's life.

Cross-reference results: the USPTO-hosted hits for "5583727" were examiner search-strategy logs and petition exhibits filed in unrelated inter partes matters (e.g. the ptacts.uspto.gov petition documents citing ("20070292720") or ("5583727") or ("6280813")... as a prior-art search string). Those are third parties citing this patent as prior art, not assignments, dockets, or assertions against it. I found nothing in RPX/Unified-type asserter coverage tying this patent number to an NPE.

---

Verdict

Insufficient data — with the practical conclusion that this patent was a non-asserted, operating-company-held asset that expired in IBM's hands.

Justification: the assignment record contains only two entries — the Parkin→IBM assignment of assignor's interest recorded 1995-07-05, and the IBM→Secretary of the Navy confirmatory license recorded 1999-01-07 (a license, not a conveyance of title) — and no assignment out of IBM thereafter through the 2015-05-15 expiry. There is no shell-entity transfer, no asserter in the chain, no cascading LLC sequence, and no litigation, which rules out every affirmative NPE signal. I stop short of a stronger positive verdict only because reel/frame numbers and the correspondents of record were not retrievable, leaving signal #3 genuinely unresolved rather than affirmatively clean; and I deliberately do not upgrade this to "Operating-company assertion," because no assertion was found and the record suggests no IBM product ever embodied claims 1 or 16.

Verify at: USPTO Assignment Center or the indexed Assignment Search — search by patent number 5583727 (or application 08/441,100). You are looking for the reel/frame of the two entries above and, most importantly, the correspondent of record on each, which is the one field this analysis could not obtain.

---

Contradiction / integrity flags versus the earlier sections

• The prior section characterizes the Navy recording as a "confirmatory license" and warns not to treat the Navy as assignee — consistent with my reading; I add that it is a government-rights encumbrance (nonexclusive license) that would travel with any future assertion of this patent.
• The prior section reports no litigation and no CAFC 2026 docket — consistent with everything in the assignment record; nothing in the chain suggests a suit was ever contemplated by an assignee.
• The prior section's status line "Expired – Fee Related (anticipated expiration 2015-05-15)" contains the same internal tension I flag above: a full-term anticipated expiration paired with a lapse-for-non-payment label. Neither section can resolve it without the maintenance-fee event history. Do not treat either the "full term" or the "fee lapse" reading as settled.
• The prior section lists "Current Assignee: International Business Machines Corp" — consistent; I found no evidence IBM ever divested this patent, including in the disk-drive divestiture.

I'll research the patent and its cited prior art. Let me start with searches on the patent itself and its key citations.

Let me research the remaining cited references.

Prior-Art Analysis for US 5,583,727 A

Scope, method, and a key distinction

Building on the earlier summary (bibliographic data, abstract, and claim breakdown are already established and not repeated here), this section addresses only the prior-art references.

Method / limitations. I worked from the authoritative Google Patents full text supplied (https://patents.google.com/patent/US5583727/en), which lists nine "Patent Citations" on the face of the patent. I ran searches to retrieve the text of those nine; several were retrievable, several were not (I flag each case). I did not search or return results for look‑alike numbers — the only other number I reference is US 5,585,986, which is expressly identified in this patent's own text as the co‑filed sibling and is not prior art; I flag it clearly.

Distinction that controls the analysis. The nine items under "Citations (9)" are references cited on the face of the patent (e.g., "cited by examiner"). Being cited is not the same as anticipating. Under 35 U.S.C. § 102 a single reference must disclose every element of a claim, arranged as claimed. The 54 entries under "Cited By" are forward citations (later patents citing '727) and are therefore not prior art at all.

None of the nine cited references — on the text available — discloses the full combination claimed in independent claim 1 or claim 16 (multi‑layer, magnetically decoupled media with differing coercivities + a write head whose field is below the first layer's coercivity + an MR read head producing discrete resistance steps from the superposed flux of both layers + sensing and logic decoding). Each is analyzed below for where it is closest and whether any dependent claim is exposed.

---

The nine cited references

Dates are taken literally from the supplied full text (note reference 7's 1985 priority with a 1995 issue, and the § 102(e) timing of references 8–9, whose filing dates precede the '727 filing date of 1995‑05‑15).

---

Per-reference analysis

2. US 5,134,533 A — Thomson‑CSF (Friederich & Creuzet) — the closest cited art

• Citation/date: US 5,134,533 A; FR priority 1989‑06‑27 (FR 8908545; FR 2 648 942 A1 pub. 1990‑12‑28); issued 1992‑07‑28. https://patents.google.com/patent/US5134533
• Description: The sensor's sensitive element is a monocrystalline magnetic metallic multilayer of magnetic layers separated by non‑magnetic layers, with antiferromagnetic‑type coupling and a transition from antiparallel to parallel alignment over a very short field interval ("spin‑flip"). When the magnetic layers are of equal thickness the multilayer "present[s] uni‑axial anisotropy in the plane of the layers"; an optional additional layer fixes magnetization. It discusses epitaxial/MBE growth on a lattice‑matched substrate.
• Potential § 102 target: This is the best candidate against the sensor subclaims (esp. claim 5, and by extension 12) — it discloses a monocrystalline, antiferromagnetically coupled multilayer exhibiting in‑plane uniaxial anisotropy with abrupt antiparallel↔parallel switching. But it does not disclose the claim‑5 requirement of at least three ferromagnetic layers in which each layer switches at a field different from the others — its teaching is a single collective spin‑flip transition, not layer‑distinct switching fields. It also lacks the media/write/decode elements of claims 1 and 16. Verdict: strong § 103 art for the sensor claims; not a clean § 102 anticipation of any claim.

6. WO 95/03604 A1 — Nonvolatile Electronics (Daughton)

• Citation/date: WO 95/03604 A1; US priority 1993‑07‑23; PCT/US94/07692 filed 1994‑07‑15; published 1995‑02‑02 (EP counterpart EP 0 710 390). https://patents.google.com/patent/WO1995003604A1
• Description: Composite films and GMR "superlattice"/"sandwich" structures of alternating ferromagnetic thin‑films separated by non‑magnetic intermediate layers; explains oscillation of interlayer coupling with spacer thickness, and lists arrangements to establish parallel vs. antiparallel states (antiferromagnetic coupling overcome by an external field; alternating high/low‑coercivity layers; exchange‑coupling alternate layers to a hard layer). Directed to MR sensors/memories.
• Potential § 102 target: Candidate against claim 5 — discloses ≥3 alternating FM/non‑magnetic layers with antiferromagnetic coupling and mechanisms for selective switching. However, it does not disclose the claim‑5 combination of uniaxial anisotropy in every layer plus layer‑distinct switching fields; its schemes rely on coercivity contrast, not strain/anisotropy‑engineered distinct switching fields. Verdict: § 103 art for claim 5 (and arguably 12/13); no clean anticipation. Timing qualifies it as prior art (published 1995‑02‑02, before the 1995‑05‑15 filing; priority 1993‑07‑23).

4. US 5,343,422 A — IBM (Kung et al.)

• Citation/date: US 5,343,422 A; filed 1993‑02‑23; issued 1994‑08‑30. https://patents.google.com/patent/[US5343422A](/patent/US5343422A)
• Description: A nonvolatile MR storage device — an array of spin‑valve MR elements (two ferromagnetic layers separated by a nonmagnetic conductor; one layer pinned, one free), with selection circuitry and sense amplifiers. This is solid‑state memory, not a moving‑media disk, and the element is two‑state, not multi‑level.
• Potential § 102 target: Relevant only as background for MR read elements and sensing circuitry. It lacks the multi‑layer media, superposed‑flux digital output, and decode logic. Verdict: no § 102 anticipation of any claim.

9. US 5,508,866 A — IBM

• Citation/date: US 5,508,866 A; filed 1994‑08‑15; issued 1996‑04‑16.
• Description: MR sensor with exchange‑coupled stabilization of a transverse bias layer.
• Potential § 102 target: Potentially background for the bias‑layer limitations of claims 15/28 ("a bias layer ... for fixing the magnetization of said nearest or farthest ferromagnetic layer"). But it concerns stabilization of a transverse bias layer, not fixing the terminal GMR layer's magnetization. Verdict: § 103/background relevance to claims 15/28; no § 102 anticipation. Timing (§ 102(e) as of 1994‑08‑15) qualifies but does not reach the claims.

8. US 5,491,600 A — IBM

• Citation/date: US 5,491,600 A; filed 1994‑05‑04; issued 1996‑02‑13.
• Description: Multi‑layer conductor lead structures in an MR head.
• Potential § 102 target: Peripheral; relates at most to the lead/connection aspects of the read head (leads 450/452 in the patent), not to any claim element that is in issue. Verdict: no § 102 anticipation.

3. US 5,231,294 A — Hitachi

• Citation/date: US 5,231,294 A; 1988‑03‑25; issued 1993‑07‑27.
• Description: Mn‑Al and Mn‑Si magnetic films and a magnetic recording medium. (Full text not retrieved in this session — flagged.)
• Potential § 102 target: At most background for the "magnetic material" limitations of claims 1/16 (media compositions with particular magnetic properties). It does not disclose the multilayer/decoupling/comparative‑coercivity or digital‑read structure. Verdict: no § 102 anticipation.

5. US 5,370,928 A — Fuji Photo Film

• Citation/date: US 5,370,928 A; 1991‑02‑15; issued 1994‑12‑06.
• Description: Magnetic recording medium. (Full text not retrieved in this session — flagged.)
• Potential § 102 target: Background on medium construction only. No § 102 anticipation.

1. US 4,975,791 A — Carlisle Memory Products Group

• Citation/date: US 4,975,791 A; 1988‑03‑22; issued 1990‑12‑04.
• Description: Recording system with controlled head‑transducer skew (magnetic recording/tape; skew control to limit crosstalk).
• Potential § 102 target: Background for a recording system with relative head‑media motion — a ubiquitous feature, already disclosed by the patent's own prior‑art discussion (FIGS. 1–2). No element of the novel combination. Verdict: no § 102 anticipation.

7. US 5,458,987 A — Nikon Corp.

• Citation/date: US 5,458,987 A; 1985‑06‑11; issued 1995‑10‑17.
• Description: Multilayered magneto‑optical recording medium — a record carrier with multiple layers, but for magneto‑optical (optical readout/erasable) purposes.
• Potential § 102 target: At most conceptually adjacent to "multiple layers on a substrate." It does not disclose magnetically decoupled, independently recordable data layers, differing coercivities, or MR readout. Verdict: no § 102 anticipation.

---

Consolidated anticipation verdict

Bottom line. On the available text, no cited reference anticipates any claim of US 5,583,727 under § 102. The two references closest to the read‑sensor core — US 5,134,533 A and WO 95/03604 A1 — disclose monocrystalline, antiferromagnetically coupled multilayers with abrupt antiparallel↔parallel transitions, which makes them the relevant § 103 art against claim 5 (and dependent claims 12–14), but neither teaches the claim‑5 requirement that each layer have a substantially uniaxial anisotropy and switch at a field different from the others — the very feature the '727 specification attributes to strain‑engineered anisotropy. The remaining seven are background/peripheral, and the independent claims 1 and 16 (which additionally require the multi‑layer decoupled media, the sub‑coercivity write scheme, and the superposed‑flux decode logic) are not anticipated by anything on the list.

---

Flags and caveats

• Non‑prior‑art sibling. US 5,585,986 ("Digital magnetoresistive sensor based on the giant magnetoresistance effect") is the co‑pending, concurrently filed application Ser. No. 08/441,133 (common specification) named in '727's own text. Same‑day‑filed, commonly owned co‑pending applications are not § 102 prior art to one another. Do not treat it as an anticipatory reference.
• Forward citations are not prior art. The 54 "Cited By" entries are later patents that cite '727; they cannot anticipate it.
• Unretrieved texts. I could not retrieve the full specifications of US 5,231,294 A, US 5,370,928 A, US 5,458,987 A, or US 5,491,600 A in this session. Their descriptions above are derived from the authoritative face‑of‑patent titles and are flagged as provisional; the anticipation verdicts for these four rest on the claim elements they are capable of reaching by their titles and are low‑confidence for that reason.
• Date fields are literal. Priority/legal‑status fields on the source page are annotated by the source as assumptions, not legal conclusions; I have reproduced them literally, including reference 7's 1985 priority.
• No litigation/CAFC activity for '727 was identified (consistent with the earlier summary's reported absence of evidence); this does not affect the § 102 analysis above.

I'll ground the § 103 analysis in the actual prior-art references cited on the face of US 5,583,727 rather than working from memory. Let me pull details on the key references.

Let me pull the remaining cited references and check the sibling application's fate.

Obviousness Analysis — U.S. Patent 5,583,727 under 35 U.S.C. § 103

Analyst note on method. This is an ex post reconstruction. I did not have access to the file wrapper (prosecution history) for App. No. 08/441,100, so I cannot say what the examiner actually applied or why the claims were allowed. What follows is a grounds-of-rejection analysis built from (a) the authoritative full text of the patent (including the applicant's own admissions); (b) the nine references appearing in the "Patent Citations" section of the Google Patents page; and (c) verification searches I ran on three of those references. I flag clearly below which references I actually verified versus which I know only from the face of the patent.

---

1. Governing law and the person of ordinary skill

Statute. Application filed 1995-05-15 and granted 1996-12-10 → pre-AIA § 103(a), with pre-AIA §§ 102(a), (b), (e), (f), (g) defining the prior-art universe. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007), and Graham v. John Deere Co., 383 U.S. 1 (1966), supply the framework (applied to pre-AIA claims in KSR itself and in In re Kahn/Arendi line).

PHOSITA. A person with an M.S. or Ph.D. in applied physics, materials science, or EE, plus 2–4 years' experience in thin-film magnetic recording heads/media, or equivalent. Such a person would be familiar as of May 1995 with: the 1988 discovery of GMR in Fe/Cr and Co/Cu multilayers; oscillatory interlayer exchange coupling (Parkin, Phys. Rev. Lett. 67, 3598 (1991), cited in the patent's own text); spin-valve sensors; buried-servo dual-layer disk media; and multilayer magneto-optical (MO) media with temperature-dependent, layer-selective writing.

The obviousness question, framed properly. The invention is a system-level combination: (1) multi-layer media with different coercivities; (2) a write scheme that writes one layer without disturbing the other; (3) a GMR read sensor whose ferromagnetic layers switch at distinct applied fields, producing a stepped (digital) resistance output; and (4) decode logic. Under KSR, the question is whether each element was known or predictable and whether there was an articulated reason to combine.

---

2. Element-by-element scope of claim 1 (the broadest claim)

Reading note (carried from the prior section): claim 1 as published contains "aid magnetic flux superposition." I read this literally as "said," and do not treat it as affecting scope.

Critical admission. The specification states:

"It would be advantageous to increase the areal data density not by reducing the size of a single magnetic bit, but by use of a plurality of independent, magnetically isolated magnetic data layers… However it is not possible to read the data from multiple magnetic data layers simultaneously, using either conventional MR sensors based on the AMR effect or proposed MR sensors based on the GMR effect, because such sensors produce an analog output signal that is not capable of directly distinguishing between the discrete magnetic field levels…" (Background)

This is a statement of the alleged problem — and it is also an admission of (i) the desire for multi-layer media, and (ii) that multi-layer media plus a read sensor was the obvious design space. The obviousness case largely reduces to whether the "analog output" premise was correct in view of the cited art. As shown below, it was not.

---

3. The cited prior art, reference by reference

3.1 US 5,134,533 A — Thomson-CSF, "Magnetoresistive sensor" (priority 1989-06-27; issued 1992-07-28) — VERIFIED

This is the single most damaging reference, and it appears twice on the patent's own record: as a front-page citation and in the Background as "U.S. Pat. No. 5,134,533." Note 5,134,533 and US5134533A are the same document (7-digit vs. 8-digit formatting) — a useful cross-check that the applicant's own background art is the examiner's primary citation.

Verified disclosures (from Google Patents / uspto.report full text):

• "the sensitive element of the sensor is constituted by a monocrystalline magnetic metallic multilayer… the multilayer being made in such a way that the layers of magnetic material present antiferromagnetic-type coupling and that the transition between the state of antiparallel alignment and that of parallel alignment takes place over a very short interval of the magnetic field."
• Substrate "must have an appropriate lattice structure so that the growth of the multilayer gives a monocrystal"; growth by MBE.
• "the layers of magnetic material are alternatively thin and thick with a ratio between the thicknesses of between 1:1 and 1:3… the transition… will be of spin-flip type (by sudden return of the magnetization of the thin layers to a given field called the spin-flip field)."
• "According to another characteristic… the layers of magnetic material are of the same thickness, the multilayer then presenting uni-axial anisotropy in the plane of the layers."
• "the magnetic metallic multilayer is covered with at least one additional layer with fixed magnetization."

URLs: https://patents.google.com/patent/US5134533 ; https://uspto.report/patent/grant/5134533

Why this matters for limitation 1e. US 5,134,533 expressly claims a GMR multilayer whose antiparallel→parallel transition occurs "over a very short interval of the magnetic field" — i.e., a step, not an analog transfer curve. That directly contradicts the '727 Background premise that GMR sensors "produce an essentially analog signal output." Further, the thin/thick "spin-flip" embodiment inherently yields at least two distinct switching fields (the thin magnetic layers reverse at the spin-flip field while the thick layers do not), i.e., a multi-level, stepped response rather than a single two-state response. (Confidence on the multi-step inference: moderate-high; I read the description and Fig. 5 is described as "the hysteresis loop of the sensor in FIG. 3," but I did not visually inspect Fig. 5.)

What US 5,134,533 does not squarely show: N independently switching layers each at its own field, and application to a multi-layer disk. Those gaps are filled by the references below.

3.2 WO 95/03604 A1 — Nonvolatile Electronics Inc. (Daughton), "Magnetic structure with stratified layers" (priority 1993-07-23; published 1995-02-02) — VERIFIED

Verified disclosures (Google Patents WO1995003604A1; EP 0710390 B1):

• Describes a superlattice of N ferromagnetic thin films separated by N−1 nonmagnetic intermediate layers exhibiting GMR/spin-valve behavior.
• "Exhibiting the giant magnetoresistive effect… requires… arrangements that permit the establishment alternatively of both parallel and antiparallel orientations of the magnetizations in the alternate ferromagnetic thin-film layers therein. One such arrangement is to have the separated ferromagnetic thin-films… antiferromagnetically coupled but to a sufficiently small degree so that the coupling field can be overcome by an external magnetic field."
• "Another arrangement is to form the ferromagnetic thin-film layers with alternating high and low coercivity materials so that the magnetization of the low coercivity material layers can be reversed without reversing the magnetizations of the others."
• "A further alternative arrangement is to provide 'soft' ferromagnetic thin-films and exchange couple every other one of them with an adjacent magnetically hard layer… so that the ferromagnetic double layer will be relatively unaffected by externally applied magnetic fields even though the magnetizations of the other ferromagnetic thin-film layers will be subject to being controlled by such an external field."

URLs: https://patents.google.com/patent/WO1995003604A1 ; https://patents.google.com/patent/EP0710390B1 ; http://patentimages.storage.googleapis.com/14/52/4f/447cb22c703368/EP0710390B1.pdf

Why this matters. WO 95/03604 supplies the express teaching that, in a GMR multilayer, individual ferromagnetic layers can be engineered to reverse at different applied fields (by Hc selection, by coupling strength, or by exchange-biasing selected layers) — which is precisely the "each ferromagnetic layer switches its magnetization… independently of the other ferromagnetic layers" mechanism recited in claims 1, 5 and 16. It thus supplies the missing teaching from US 5,134,533 and, importantly, does so with an express stated purpose (multiple distinguishable magnetic states in one stack).

3.3 US 5,343,422 A — IBM (Kung et al.), "Nonvolatile magnetoresistive storage device using spin valve effect" (filed 1993-02-23; issued 1994-08-30) — VERIFIED

Verified: a GMR/spin-valve element with two ferromagnetic layers separated by a nonmagnetic metallic (Cu) layer; one layer's magnetization "fixed" (by exchange coupling to an antiferromagnetic layer or by "a hard magnetic material having a greater coercivity than… said other layer"); the other layer free to switch parallel/antiparallel; a sense-current/voltage detection and array logic architecture; and the express statement that magnetization states can be fixed "provided layer 28 is of a sufficiently hard magnetic material… or has sufficiently high anisotropy to retain its magnetization during state switching."

URL: https://patents.google.com/patent/[US5343422A](/patent/US5343422A)/en

Why this matters. It supplies (i) the Cu-spacer spin-valve stack, (ii) the use of differing coercivity/anisotropy between layers to make one layer switch and another not, and (iii) MR element + sensing circuitry + selection/decode logic — i.e., claim 1's elements (1g) in a nonvolatile memory context. The '727 specification itself acknowledges the sensor "also has application to other technologies, such as… magnetic random access memory systems wherein a magnetoresistive element serves as a bit cell," confirming the art-recognized equivalence of the MRAM and MR-head contexts.

3.4 US 5,458,987 A — Nikon, "Multilayered magnetooptical recording medium" (priority 1985-06-11; issued 1995-10-17) — cited of record; disclosure not independently verified

From the face of the patent only. Title and the related Nikon-family documents I did retrieve (US 5,164,926; US 5,164,925; US 5,235,569; US 4,910,622; EP 0 384 746) establish the relevant genus of art with high confidence: two magnetic layers of different coercivity and different Curie/compensation temperatures, exchange-coupled, in which a laser beam locally heats the medium so that a selected layer's Hc drops below the applied field while the other layer's does not — with explicit formulas for "reversing the direction of magnetization of the second layer by the first external field He1" without reversing the first layer.

Why this matters. This is the art from which the '727 specification's selectable layer write scheme (FIG. 8A/8B; Tb–Fe–Co; compensation temperature Tcomp above/below Toper) is drawn, and it motivates claim 17's heater (and dependent claims 2/18's buried-servo write scheme in its broadest sense). The '727 text reads as a transplantation of MO layer-selective thermomagnetic writing into a longitudinal disk.

3.5 Remaining face citations — not independently verified

Common-ownership caveat. US 5,343,422, US 5,491,600 and US 5,508,866 are IBM documents. If this ever went to a reexamination or validity challenge, expect an argument under pre-AIA § 103(c) that commonly owned § 102(e)/(f)/(g) art cannot be used for obviousness. Note that § 103(c)'s commonly-owned exception as amended in 1999 was keyed to patents granted on or after the 1999 amendment date; the '727 patent granted 1996-12-10, so the exception's applicability is genuinely contestable. Flag: I have not researched this point to a conclusion and it should be checked before any substantive filing.

3.6 One point worth raising even though it is not in the "Citations (9)" table

The Google Patents page lists US 5,844,755 A (Oki Electric, "Giant magnetoresistive information recording medium, and associated recording and reproducing method and apparatus") under "Cited By," with a priority date of 1994-12-22 — i.e., before the '727 filing date of 1995-05-15. If its underlying U.S. application was filed on that date, it would be pre-AIA § 102(e) prior art (granted 1998-12-01). Its title alone — GMR information recording medium plus a recording and reproducing method and apparatus — makes it potentially highly material to claims 1 and 16. I could not retrieve its disclosure (search budget exhausted). This is the single most important verification gap in this analysis and should be closed first.

Also note the applicant's own Background cites K. Inomata et al., J. Appl. Phys. 74(6), 15 Sep 1993, describing GMR in single-crystalline multilayers exhibiting uniaxial magnetic anisotropy. Dated >1 year before the 1995-05-15 filing, that publication is § 102(b) art in its own right — and it is a printed publication, so the common-inventorship/§ 103(c) escape does not apply to it. Likewise the patent's own cited Harp & Parkin, Appl. Phys. Lett. 65(24), 3063 (12 Dec. 1994) is § 102(a)/(b) art as to the sputter-deposited crystalline multilayer and the Fe/Pt-on-MgO(110) seed stack (relevant to claims 6–9, 22–24).

---

4. Proposed grounds of rejection

Ground 1 — Claim 1 obvious over US 5,134,533 in view of WO 95/03604, further in view of US 5,343,422 (and the buried-servo media admitted as known)

Articulated reason to combine (KSR). (i) Both references are in the same field and address the same device genus (GMR multilayers) — the "familiar element" rationale. (ii) WO 95/03604 expressly frames the problem of obtaining distinguishable magnetic states within a single superlattice and provides the solution (layer-to-layer coercivity/coupling differentiation), which is a known technique to improve a similar device. (iii) The motivation for combining with multi-layer media is stated by the applicant itself in the Background (increase areal density without shrinking the bit). (iv) There is a reasonable expectation of success: US 5,134,533 already teaches the step-like transition and two-field switching, and WO 95/03604 teaches how to extend that to N differentiable states.

The design-choice point (KSR, Part II). Even absent WO 95/03604's express teaching, differentiating switching fields among layers of a GMR stack would have been an obvious result-effective variable optimization: Hswitch is set by Hki versus the sum of the antiferromagnetic coupling fields — as the '727 specification itself explains. Varying anisotropy by strain (which relaxes with distance from the seed layer), varying spacer thickness (which tunes coupling), and varying layer thickness/composition were all known, routine knobs for a PHOSITA working with MBE/sputtered superlattices. In re Applied Materials / KSR ("a court must ask whether the improvement is more than the predictable use of prior art elements according to their established functions").

Applicant's expected rebuttal — and its weakness. The '727 Background asserts it was "not possible to read… simultaneously" because GMR sensors give "an essentially analog signal output." The cited US 5,134,533 says the opposite about its own device (transition "over a very short interval of the magnetic field"). An assertion that the art could not work, contradicted by a reference of record, is not a persuasive teaching-away.

Ground 2 — Claim 16 obvious over US 5,343,422 in view of US 5,134,533 and WO 95/03604, further in view of a disk-drive art reference (e.g., US 4,167,765 / US 4,975,791) for the motor, carrier, actuator and supporting-means elements

Claim 16 recites a structural species: disk + motor + inductive write head + GMR sensor with permalloy layers and Cu spacer layers, single-crystalline face, seed layer, ≥3 permalloy layers antiferromagnetically coupled, and differing uniaxial anisotropy per layer. Each element:

• Permalloy + Cu — US 5,134,533 and US 5,343,422 (Cu spacer; NiFe-type ferromagnets).
• Essentially single-crystalline face / MgO(110) / Fe–Pt seed — US 5,134,533 (monocrystalline, MBE, lattice-matched substrate) plus the applicant's own cited Harp & Parkin publication (Fe/Pt seed on MgO(110) by sputtering).
• Differing uniaxial anisotropy per layer — WO 95/03604's differentiation teaching; the strain-relaxation mechanism is an inherent consequence of epitaxial growth on a lattice-mismatched substrate.
• Motor / carrier / actuator / supporting means — admittedly conventional (the patent's own FIGS. 1–2 and the Watrous suspension reference, US 4,167,765).

The claim 16 combination is therefore a predictable arrangement of known structural elements, each performing its known function.

Ground 3 — Claim 17 (heater) obvious over the Ground-2 combination in view of US 5,458,987 (Nikon) or the Nikon/Canon MO family

Claim 17 adds only "a heater for heating the data layers prior to writing." The Nikon multilayered-MO art cited of record — and the broader Nikon/Canon thermomagnetic double-layer art I retrieved (US 5,164,926, "recording is enabled… if a bias field (Hb) capable of cancelling a decreased Hc is applied to a magnetic material having the decreased Hc at a predetermined temperature higher than a room temperature"; US 5,164,925; US 4,910,622; EP 0 384 746) — expressly discloses laser/local heating of a two-layer magnetic medium to selectively write one layer based on differing coercivity-temperature dependence. Combining that known heat-assisted, layer-selective write technique with a multi-layer disk write head is a straightforward application of a known technique to a similar device, with predictable results. The patent's own FIG. 8A/8B discussion (Tb–Fe–Co; Tcomp) is drawn directly from that art.

Ground 4 — Claims 12, 13, 14, 25, 26, 27 obvious as design choices / result-effective variables over the Ground-1/2 combinations

These claims recite that each layer has a different anisotropy (12, 25), that each neighboring pair has a different coupling field strength (13, 26), and that each spacer layer has a different thickness (14, 27). The '727 specification itself recites the full menu of ways to achieve these differences (layer thickness ti; magnetization Mi; spacer thickness; composition; morphology via pressure/rate/temperature/ion bombardment/annealing) — i.e., the claim recites an objective to be achieved by a set of expressly known, routine optimization parameters. Where the reference teaches the desirability of per-layer differentiation (WO 95/03604), optimizing the known parameter to achieve it is obvious.

Ground 5 — Claims 6, 7, 9, 22, 23, 24 (MgO(110), 110 crystalline face, Fe/Pt seed) obvious over the Ground-2 combination in view of Harp & Parkin (1994)

The applicant's own citation, Harp and Parkin, Appl. Phys. Lett. 65(24), 3063 (12 Dec. 1994), discloses sputter-deposited crystalline multilayers, and the '727 specification identifies the preferred substrate as polished MgO(110) with a 5 Å Fe / 30–50 Å Pt bilayer seed grown at ≥500 °C. A printed publication is prior art regardless of common inventorship, so these structural claims face § 102(a)/(b)/§ 103 art originating from the inventor's own group.

Claims 3, 4, 20, 21 (longitudinal vs. perpendicular domains)

Alternative, conventional recording modes explicitly declared interchangeable in the art and in the patent's own text ("the invention is also applicable to vertical or perpendicular recording"). Obvious.

Claims 5, 8, 10, 11, 15, 28 (sensor architecture details)

• ≥3 antiferromagnetically coupled FM layers with uniaxial anisotropy and layer-specific switching fields (5) — US 5,134,533 + WO 95/03604.
• Seed layer (8) — US 5,134,533 (lattice-matched growth) + Harp & Parkin.
• Permalloy layers (10); Cu spacers (11) — US 5,134,533; US 5,343,422.
• Bias layer fixing nearest/farthest layer (15, 28) — US 5,343,422 ("magnetization direction of magnetic layer 28 is fixed… by exchange coupling with an antiferromagnetic layer 30") and US 5,508,866 (exchange-coupled stabilization). Both of claims 15/28's alternatives — "magnetization fixed by a separate biasing layer" — are expressly disclosed in US 5,134,533 ("the magnetic metallic multilayer is covered with at least one additional layer with fixed magnetization").

Claims 2, 18 (pre-written servo data in the first data layer)

The specification concedes: "A two layer magnetic recording disk with the bottom higher coercivity magnetic layer containing servo data is known, and referred to as a 'buried servo' implementation, but the reading of the servo data is done by a separate dedicated servo head." The only difference from the admitted art is that the '727 reads the servo layer with the same head as the user layer — which is the very feature supplied by the Ground-1 combination. Claims 2 and 18 are probably among the most vulnerable in the patent.

---

5. Where the nonobviousness case is strongest (and what would have to be argued)

1. The system-level integration. No single reference, and arguably no combination of the nine face citations, expressly discloses a disk drive that (i) writes selectable layers of a decoupled (not exchange-coupled) dual-layer medium using coercivity-temperature differentiation, and (ii) reads both layers simultaneously through one GMR stack whose output is decoded into two separate digital data streams. Ground 1 assembles this from four references plus an admission. That is a legitimate § 103 case under KSR, but it is a combination case and will turn on the strength of the articulated motivation.
2. "Analog vs. digital" as an asserted unexpected result. The applicant could argue the stepped-resistance output (FIGS. 6A/9C) was an unexpected property of a crystalline strain-graded multilayer. This is weakened by (a) US 5,134,533's "very short interval" transition and (b) the absence of any comparative data in the specification showing that the result would have been unexpected rather than merely designed-for. No secondary-considerations evidence (commercial success, licensing, industry praise, long-felt need) appears in the specification. The 54 forward citations (Seagate, Hitachi GST, etc.) show technological significance, but forward citation alone is weak evidence of nonobviousness; there is no established nexus to a commercial embodiment.
3. The exchange-coupled vs. decoupled distinction. The MO art (Nikon) that motivates the heater uses exchange-coupled layers (a wall energy σw dominates), whereas claims 1/16 require a decoupling layer. A patentee could argue that transferring the MO layer-selective writing scheme to a magnetically decoupled pair is not a simple substitution. This is the most technically respectable rebuttal, but it is a design argument, not a teaching-away; the patent itself supplies a decoupling-layer material set (Al2O3, Ta, Cr, Re) drawn from known spacer technology.

---

6. Vulnerability ranking of the claims

Confidence: I would rate the Ground-1/Ground-2 case against claims 1 and 16 as moderate-to-strong on the sensor side (US 5,134,533 + WO 95/03604 + US 5,343,422 are a genuinely tight trio) and moderate on the media/write-scheme side (which rests significantly on the applicant's own admissions). Claims 2/18 (buried servo) and 17 (heater) are individually the cleanest rejections.

---

7. Explicit uncertainty flags

1. No file wrapper. I could not retrieve the prosecution history of App. No. 08/441,100, so I do not know which of these references the examiner actually applied or what arguments were overcome. This analysis is a reconstruction, not a record of what happened.
2. Unverified reference disclosures. I verified US 5,134,533, WO 95/03604 (and its EP 0710390 counterpart), and US 5,343,422 from live sources. I did not verify US 4,975,791, US 5,231,294, US 5,370,928, US 5,458,987, US 5,491,600, or US 5,508,866; statements about those rest on their titles and on the related Nikon/Canon MO family documents I did retrieve. Do not quote my characterization of those six to a court or client without checking the primary documents.
3. The Oki gap. US 5,844,755 (Oki Electric; priority 1994-12-22; issued 1998-12-01) has a priority date preceding the '727 filing date and is titled "Giant magnetoresistive information recording medium, and associated recording and reproducing method and apparatus." If its U.S. filing date is also pre-1995-05-15, it is potential § 102(e) art and could be a more material reference than anything I have analyzed. I could not retrieve its content. This should be the first item verified.
4. § 103(c) common-ownership question (note 3.5) is unresolved and may materially affect whether the three IBM references can be used for obviousness.
5. Date arithmetic. Today's date is given as 2026-09-29 in the system context and April 26, 2026 in the task prompt. Nothing in this analysis turns on the discrepancy, but I flag it per the operating rule against auto-correction.
6. The '727 patent expired 2015-05-15 (anticipated expiration per Google Patents), consistent with the prior litigation section's finding of no identified litigation. This analysis is therefore of historical/validity interest only — it has no live enforcement consequence.