US 9897486

I'll search for litigation/docket information on this patent number, then summarize from the authoritative patent text I have.

US Patent 9,897,486 — Summary

Important scope note up front: I ran targeted searches against USPTO/Google Patents records and for CAFC 2026 dockets, ITC §337 proceedings, and district-court complaints referencing this number. I found no CAFC 2026 docket, ITC investigation, or district-court litigation naming US 9,897,486. I also want to flag a potential confusion: one search surfaced a 2026 ITC semiconductor dispute (Infineon v. Innoscience, Inv. No. 337-TA-1414) involving U.S. Patent No. 9,899,481 — that is a different patent (a power-semiconductor patent), not 9,897,486, and per your instruction I am not treating it as a match. Treat "no litigation found" as a negative search result, not proof of absence.

The bibliographic and technical details below come from the authoritative full patent text.

Bibliographic data

Abstract (verbatim)

A method of calibrating a measuring apparatus includes determining apparatus parameters that have an influence on a measurement spectrum generated by the measuring apparatus, generating the measurement spectrum by exposing a measurement target on a sample to light generated by the measuring apparatus, calculating an error of the apparatus parameters by comparing the measurement spectrum to an ideal spectrum corresponding to the apparatus parameters, and calibrating the measuring apparatus based on the calculated error of the apparatus parameters.

Plain-language technical overview

The patent addresses error correction in spectroscopic measurement tools (particularly spectroscopic ellipsometry, SE) used in semiconductor production. The core problem: multiple SE tools in a fab drift from one another, and traditional "spectral error-based calibration" (SEBC) relies on repeatedly measuring the same reference sample at each process step (potentially 100+ corrections per product) while only aligning tool averages, not tool tendencies — an inefficient and imperfect fix.

The patent's approach:

1. Select only the apparatus parameters with a substantial spectral influence. Rather than treating every tool element as a variable, the disclosure determines that the angle of incidence (AOI) and the sensor/array-detector alignment state dominate spectral distortion, while light-source wavelength/NA, polarizer angle, and analyzer angle are stable and can be ignored. This selectivity keeps the calibration math tractable.
2. Measure a calibration sample (e.g., a bare wafer) and generate a measurement spectrum.
3. Compare the measurement spectrum to an "ideal" modeled spectrum (corresponding to the parameters in their normal/reference state, e.g., AOI = 70°, sensor at reference position). Spectrum modeling iterates the apparatus-parameter values until the model matches the measurement, and the discrepancy is quantified as a numeric error (examples: AOI off by 1°, sensor alignment off by 0.5 mm).
4. Calibrate via an inverse operation. The quantified errors are expressed as a matrix; the measurement spectrum (a row matrix) is multiplied by the inverse matrix to produce a calibrated spectrum. This corrects the tool's intrinsic error once, so calibration need not be repeated for every subsequent device or process step.
5. Then measure actual devices by producing a calibrated spectrum and fitting a model spectrum (varying sample/device parameters, not apparatus parameters) to extract thickness or 3D shape.

The specification notes the invention extends to spectroscopic reflectometry (SR), XPS, XRR, and UPS tools, and to measuring semiconductor devices, wafers, LCD panels, and TFT panels. It also describes cascaded use across multiple process steps with one or multiple tools.

Independent claims — plain-language overview

Uncertainty flag: The authoritative text I retrieved contains the abstract, summary, and detailed description but not the literal claim set. The summary section tracks the claim language closely and lists three independent-claim families, so the overview below reflects that summary language rather than verbatim claim text. I cannot confirm exact claim numbering, dependencies, or any narrowing limitations added during prosecution without the granted claims in front of me.

1. Method of calibrating a measuring apparatus — Determine the apparatus parameters that influence a measurement spectrum generated by the apparatus; generate the measurement spectrum by exposing a measurement target on a sample to the apparatus's light; calculate an error of those apparatus parameters by comparing the measurement spectrum to an ideal spectrum corresponding to the (reference-state) apparatus parameters; and calibrate the apparatus based on the calculated error.

2. Method of measuring a characteristic of a device — Calibrate the measuring apparatus based on apparatus parameters influencing a first measurement spectrum (where the spectrum is generated by exposing a measurement target on the device to the apparatus's light); determine a calibrated spectrum for the device using the calibrated apparatus; determine a first approximate spectrum by changing values of device-related parameters during spectrum modeling until it is substantially consistent with the calibrated spectrum; and acquire information about the device characteristic (e.g., thickness or shape) from the parameter values used to obtain that approximate spectrum.

3. Method of manufacturing a device — The same sequence as claim family 2, but with a semiconductor process related to a measurement being performed on the device between calibration and the calibrated-spectrum determination; information about the device characteristic is then acquired from the fitted device parameters. The stated benefit is fewer product defects and higher yield through improved measurement reliability.

Bottom line

US 9,897,486 is a Samsung Electronics patent on one-time, model-based, select-parameter calibration of spectroscopic measurement tools — essentially computing a tool's intrinsic error against an ideal spectrum and inverting it, so that a single calibration serves all later measurements. I found no CAFC 2026 or other litigation activity for this number, and I deliberately excluded the similarly numbered but unrelated US 9,899,481. The claim overview is based on the patent's summary rather than the verbatim granted claims, which is the one part of this answer I'd want to verify against the issued claim set before relying on it.

I'll research litigation involving US Patent 9897486 across patent litigation trackers and court sources.

The initial results reference different patents (e.g., 9,909,748, the "748 Patent"). Let me search more specifically for the exact patent number and litigation trackers.

Let me search patent litigation databases and trackers directly.

Summary of Findings

No known litigation involving U.S. Patent No. 9,897,486 was identified in the sources searched.

What I searched and what came back

I ran targeted queries against patent-litigation-oriented sources (including PTAB/USPTO PTACTS filings, court dockets surfaced via CourtListener, and general litigation trackers) using the exact identifiers "9897486", "US9897486," and "9,897,486."

Every result that appeared was a false positive for a different patent number, and I am expressly not reporting them as litigation for '486:

None of these involve U.S. Patent No. 9,897,486, and per your instruction I have not substituted any "similar number" results.

About the patent itself (context)

US 9,897,486 B2, titled "Method of calibrating and using a measuring apparatus that performs measurements using a spectrum of light," is owned by Samsung Electronics Co., Ltd. It issued February 20, 2018, from application 15/402,751 (filed Jan. 10, 2017), claiming priority to Korean Application 10-2016-0059789 (filed May 16, 2016). It is presently listed as Active with an anticipated expiration of January 10, 2037. Its subject matter is spectroscopic-ellipsometry calibration — generally not the type of patent that appears in the public NPE assertion campaigns that generate most reported litigation.

Important caveats on confidence

• I could not confirm the absence of litigation with high confidence. My searches returned largely irrelevant noise, and I did not obtain a clean, authoritative "no cases" confirmation from a primary litigation database (e.g., Docket Navigator, Lex Machina, Unified Patents caselist, or a PACER/CAFC docket search) that indexes by patent number.
• The sources that would be definitive — Unified Patents litigation portal, PACER, and Docket Navigator — were not successfully retrieved for this exact patent number.
• A patent can be asserted in a district court case, an ITC Section 337 action, or an IPR/PGR that does not surface in open web search, especially for a Samsung-held process/metrology patent.

Recommended next steps to confirm

To reach a definitive determination, please run the patent number directly in:

1. Unified Patents litigation caselist — https://portal.unifiedpatents.com/litigation/caselist (search "9897486")
2. USPTO PTAB (Patent Trial & Appeal Board) — search for IPR/PGR involving U.S. 9,897,486
3. PACER / CourtListener RECAP and Docket Navigator / Lex Machina — patent-number indexed dockets
4. USITC EDIS — for any Section 337 investigations

Bottom line: Based on the searches performed, I found no litigation involving U.S. Patent No. 9,897,486, and I decline to attribute any of the analogous-number results to it. However, because I could not query the authoritative patent-litigation databases directly, I cannot state with high confidence that no litigation exists — only that none was located.

I'll verify the canonical "no proceedings" result with targeted searches before writing this up.

Both searches returned only false positives on other patent numbers. Let me run two more variants before finalizing.

Proceedings overview

Total AIA trial proceedings on file for US 9,897,486: zero (0). Breakdown: 0 active, 0 with claims invalidated, 0 with claims sustained, 0 settled, 0 institution denials. The USPTO Open Data Portal (canonical source for the structured block in this prompt) returns no IPR, PGR, or CBM proceeding naming this patent, and four targeted web searches (patent number in two formats, plus "inter partes review Samsung" and a docket-tracker query) surfaced only false positives on other patent numbers — the same 9,909,748 / 8,301,946 / analog-number noise already documented in the litigation section above. I did not find a single filing, decision, or docket entry indexed to this patent.

Defensive posture: zero. This is not a "hardened patent survived two IPRs" situation and it is not a "claims 1–5 are canceled" situation — it is simply an untested patent. That cuts both ways for a defendant:

• In your favor: the patent has never been before the Board, so no claim has been narrowed by adverse judgment, no prosecution disclaimer has been locked in by a patent owner's IPR arguments, and no estoppel has run against anyone. The full prior-art universe is available to you.
• Against you: you have no PTAB roadmap — no panel reasoning, no adopted claim constructions, no expert findings — to lean on. Every validity position must be built from scratch.

---

Proceedings list

No proceedings to enumerate. There is no FWD to quote, no institution decision to characterize, no judge panel to identify, and no appeal to trace for this patent number. Per the task constraints, I will not invent proceeding numbers, panels, or claim-level outcomes to fill this section.

---

Strategic summary

Claim status — all claims are UNTESTED. Because no AIA trial was ever instituted, there is no PTAB record of any claim of 9,897,486 being canceled, no claim being held patentable, and no claim being narrowed by a certificate of correction or adverse judgment. I flag a related uncertainty carried over from the patent summary: the authoritative text I retrieved did not include the verbatim granted claim set, so I cannot state claim numbers or the independent/dependent structure with confidence. Before relying on any specific claim, pull the granted claims from the USPTO Patent Center / Google Patents. What I can say from the specification's summary language is that the patent describes three independent-claim families (a calibration method, a device-characterization method, and a device-manufacturing method), not a large claim set.

Estoppel landscape — clean, and that is the practical headline. Section 315(e)(2) estoppel only attaches to a petitioner who obtains a final written decision. Since there is no FWD here, there is no IPR estoppel bar on any party. You may raise any § 102/§ 103 ground in district court or the ITC, including grounds you would otherwise have had to bring (or lose) at the PTAB. There is also no IPR-based "roadmap estoppel" risk — no petitioner has already litigated and lost a ground you might want to repackage. The full scope of prior art against spectroscopic-ellipsometry calibration — SEBC references, SE tool-calibration literature, and the applicant's own cited art — remains open.

Pattern signals. None. No repeat-petitioner pattern (no petitioner has filed even once), no patent-owner PTAB-appeal history, and no defensive aggregator (e.g., Unified Patents) in the chain. For context, the search noise did show Samsung as one of the most prolific IPR petitioners overall (the briefs cite figures like 1,130 Samsung-filed IPR petitions through 2024), but that is irrelevant to this patent — here Samsung is the owner, not a challenger, and it has not used the Board offensively against this number.

Why the silence is plausible: this is a process/metrology patent owned by a large operating company, not an NPE assertion asset. Patents of this type are typically cross-licensed rather than asserted, which is consistent with (though not proof of) the absence of both litigation and IPRs. Note the cross-reference to the litigation section: no CAFC 2026 docket, no ITC § 337 action, and no district-court complaint naming this number was located either — so the "no PTAB" result is not an outlier, it matches the absence of any assertion activity.

Caveat on confidence: The ODP structured block is authoritative and says "no proceedings on file," and web search corroborates it. But I could not directly query PTAB E2E / PTAB Center or Docket Navigator by patent number, and a very recently filed petition (pre-ODP ingest) can lag. Treat this as "none located and none on file per the canonical source," not as forensic proof of absence.

---

Recommended next steps

1. You are not facing a claim-invalidation defense, so don't look for one. There is no FWD to link or quote, and no canceled claim to hang a 101/112 or § 285 argument on. Any invalidity theory must start fresh.
2. Search the record yourself before relying on this. Run the patent number in:
   • USPTO PTAB Center / PTAB E2E — https://ptacts.uspto.gov/ptacts/ (search by patent number 9,897,486) — also the location for any FWD, institution decision, or E2E exhibit if one later appears.
   • USPTO PTAB Decisions — https://www.uspto.gov/patents/ptab/decisions
   • CourtListener — https://www.courtlistener.com/ (for any CAFC appeal that would presuppose an FWD)
   • Unified Patents litigation portal — https://portal.unifiedpatents.com/litigation/caselist
3. If an IPR is filed while you are in litigation, watch the statutory clock. Under 35 U.S.C. § 316(a)(11) the Board must issue a final determination within one year of institution (extendable up to six months for good cause). Milestones to diary: petition filing → PO preliminary response (3 months) → institution decision deadline (3 months from PO's preliminary response or filing date) → FWD (1 year from institution). Those are the dates that drive a stay motion.
4. Because there is no PTAB activity, the absence is itself the signal — this patent has not been asserted or challenged in any indexed venue I could find, so a demand letter invoking it should be pressure-tested for whether the patent is actually the value driver of the portfolio, rather than treated as a battle-hardened asset.

One cross-reference flag: the prior litigation section concluded "no litigation found" with the same confidence caveat and the same false-positive set (IPR2021-00663 / Ancora on U.S. 9,909,748; the Lenovo "946 patent"; ACQIS v. ASUS; the Chinese ZL '486.2). This PTAB section reaches the same "no activity" bottom line by independent means, so the two sections are consistent — but both share the identical limitation that the authoritative litigation databases were not directly retrievable for this number.

I'll research the assignment record for US 9,897,486 before reconstructing the chain.

Let me try more targeted searches for the assignment record and any ownership changes.

Let me attempt to retrieve the actual assignment record metadata and check assignee status.

Research limitation — read first

I attempted to pull the live Assignment Center record (reel/frame, correspondent of record) via several targeted queries against USPTO Assignment Search, Google Patents legal events, and general web sources. The searches returned unrelated patent noise, and I was unable to retrieve the reel/frame number or the recorded correspondent for the single assignment on this patent. Per your constraints, I will not fabricate those fields — I mark them as not retrieved below. Everything else is grounded in the authoritative patent text's Google Patents legal-events block, which I do have.

Note also the prior sections' finding: no litigation, PTAB, or ITC activity was located for US 9,897,486 (with the caveat that authoritative litigation databases were not successfully queried). That negative result is carried forward here.

---

Inventors

Unusual-pattern check: No evidence of the "all inventors depart within 12 months of filing" precursor to a fire-sale. The inventors appear in the initial assignment to Samsung only; there is no inventor assignor appearing in any later conveyance (there are no later conveyances). The application was a US national-stage-style direct filing with a Korean priority claim (KR 10-2016-0059789, filed 2016-05-16), typical of Samsung's Korean-origin portfolio, not an inventor-owned startup filing.

---

Original assignee

Samsung Electronics Co., Ltd. (Suwon-si, Gyeonggi-do, Republic of Korea) — named as both Original Assignee and Current Assignee on the Google Patents record (with Google's standard disclaimer that it has performed no legal analysis of the assignee list).

• Primary line of business: Consumer electronics and semiconductor manufacturing — including the semiconductor fabrication and metrology operations that this patent's spectroscopic-ellipsometry (SE) calibration techniques are designed to serve.
• Product embodying the claims: The claimed subject matter is a method (calibration + measurement + device-manufacturing methods), not a device. Samsung does not appear to sell an SE metrology tool; it practices the method internally in its own fabs and metrology flows. Whether any specific product embodies the claims is not determinable from the record.
• Current status: Operating. Samsung Electronics is one of the world's largest publicly traded companies (KRX: 005930), not acquired, dissolved, or in bankruptcy.

---

Assignment timeline

The patent's legal-events record shows exactly one recorded assignment — the initial inventor→corporate assignment. No post-issuance transfers appear.

• 2017-01-10 (executed) / recorded 2017-01-10 — Reel NNNNNN/NNNN (not retrieved — see limitation note)
  • Conveyance: Assignment — "ASSIGNMENT OF ASSIGNORS' INTEREST (SEE DOCUMENT FOR DETAILS)". The Google Patents record labels this event "Assigned to SAMSUNG ELECTRONICS CO., LTD. — reassignment."
  • Assignor(s): CHOI, Chang-Hoon; KIM, Jong-Jeong; JO, Tae-Yong; LEE, Young-Joo (all four named inventors)
  • Assignee: Samsung Electronics Co., Ltd.
  • Correspondent: Not retrieved. Could not obtain the recording attorney/agent of record from the sources queried. I therefore cannot run the repeat-correspondent test.
  • Context: Initial assignment to the operating parent — standard employment/obligation-to-assign transfer filed concurrently with the US application (filing date 2017-01-10). Not a fire-sale, reorg, securitization, or transfer-to-asserter.

No other events (no security agreement, no merger, no change of name, no license, no release, no correction) appear on the record.

Downstream legal events (not assignments, for completeness): US 2017/0328771 A1 published 2017-11-16; patent issued 2018-02-20; status Active; anticipated expiration 2037-01-10 (Google Patents states this as an assumption).

---

Timeline diagram

timeline
    title Ownership of US 9897486
    2016 : Korean priority application filed
    2017 : US application 15 402 751 filed by Samsung
         : Inventors assign rights to Samsung Electronics
    2018 : Patent issued to Samsung
    2037 : Anticipated expiration per Google Patents

---

NPE / troll-pattern signals

---

Verdict

Insufficient data (only the original assignment is on record).

Substantively, the record is as clean as it gets: a single, filing-date assignment from the four inventors to their operating employer, Samsung Electronics Co., Ltd., with no downstream conveyances, no shell entity, no known asserter, no cascading or pre-litigation transfer, and no bankruptcy or privateering event. Every NPE signal that could be assessed is not present, and Samsung is an operating company that practices the method internally. The verdict is "insufficient data" rather than "defensive/non-asserting" only because (a) the rubric maps "only the original assignment" to that category, and (b) I could not retrieve the reel/frame or correspondent of record — the two fields that would let me close the repeat-correspondent test and definitively confirm there are no later, lightly-indexed recordings.

Verification link: USPTO Assignment Center search — https://assignmentcenter.uspto.gov/ (search patent number 9897486); mirror index at https://assignment.uspto.gov/patent/index.html. Google Patents legal events: https://patents.google.com/patent/US9897486/en

To close the two open fields, run the patent number directly in the Assignment Center and capture: (1) the reel/frame of the 2017-01-10 assignment, and (2) the correspondent attorney/firm of record — then compare that correspondent against the firm used on Samsung's other filings to confirm no anomalous recorder is touching this number.

I'll search for the patent record and its cited prior art references across USPTO/Google Patents sources.

Let me try more targeted queries to surface the patent's own citation list.

The citation lists aren't surfacing directly. Let me try patent database mirrors that display "References Cited" tables.

Bottom line up front

I could not retrieve the "Patent Citations" (backward/face-of-patent "References Cited") list for US 9,897,486 B2 within this session. That is the list your task is asking me to analyze — the examiner-cited and applicant-cited references that serve as the § 102/§ 103 prior-art candidates against '486.

Because I do not have that list, I cannot name specific references or map them to specific claims without fabricating. I am not going to invent citations, dates, or § 102 mappings. Below I give (1) exactly what I searched, (2) the two verified data points I did find — both forward citations, not prior art against '486 — (3) the false positives I encountered and rejected, and (4) the precise places to pull the authoritative list.

---

What I searched (all queries returned noise or unrelated records)

Two notes on the authoritative text you supplied: it contains the abstract, summary, and detailed description, but the front-page "(56) References Cited" section and the granted claim set are both absent. That is the root cause of the gap — I cannot analyze a citation list that isn't in the source material and isn't retrievable by search.

---

Verified data points (both are FORWARD citations — not § 102 prior art against '486)

These are documents that cite US 9,897,486; they are relevant to infringement/validity landscape work but are not prior art against '486 and cannot "anticipate" its claims:

1. US 2006/0238759 A1 / US 9,188,543 B2 — "Spectroscopic polarimetry" / "Inspection device and method thereof," General Electric Company, priority 2011-04-18, granted 2015-11-17. On that patent's Google Patents page, US 9,897,486 B2 appears in the "Cited By (14)" table.

   • URL: https://patents.google.com/patent/US20060238759
   • Field relevance: spectroscopic polarimetry / inspection — same measurement family as '486, hence the citation.

2. KR 10-2014-0056261 A — "Multi-analyzer angle spectroscopic ellipsometry" (spectrometer movable among predetermined discrete angular positions) — its "Cited By (2)" table lists US 9,897,486 B2.

   • URL: https://patents.google.com/patent/KR20140056261A/en

3. US 7,002,693 B1 — "Thickness measurement method and apparatus" — its "Cited By (2)" table also lists US 9,897,486 B2.

   • URL: https://patents.google.com/patent/US7002693

Caveat: I could only confirm these as appearing in "Cited By" tables. I could not open the '486 page to confirm the count or the complete forward-citation set (the "Cited By (14)" figure came from a different patent's page, so it may not equal '486's own forward-citation count).

---

False positives I encountered and expressly rejected

Per your standing rule, I did not auto-correct or substitute any of these similar-looking numbers into the '486 record.

---

What is genuinely in the file (grounded in the authoritative text)

The only "prior art" discussion that the supplied '486 text itself contains is the comparative/background technique the inventors describe and position the invention against:

• Spectral error-based calibration (SEBC) — the conventional approach in which, "to correct errors between measuring apparatuses, a specific sample is measured by each of the measuring apparatuses, and correction is performed by adding or subtracting an error to or from a measured result, based on any one measuring apparatus," applied at every process step (potentially "100 or more" times per product). The specification characterizes it as matching only averages, so "tendencies in the measuring apparatuses do not coincide."

This is background art as described by the applicant, not an examiner citation, and the text does not identify a specific SEBC reference document (patent or NPL) by number or author. So it cannot be cited as a § 102 reference by identifier without fabrication.

---

Why I cannot supply a § 102 claim mapping

To say a reference "potentially anticipates claim X under 35 U.S.C. § 102," I need two things, and I have neither:

1. The reference list — not retrieved (see above).
2. The granted claim text of US 9,897,486 — the authoritative text you provided stops at the detailed description and does not include the claims. The previously generated summary section flagged this same gap for the claim overview. So any statement of the form "Reference A anticipates claim 1" would require me to reconstruct claim 1 from the Summary-of-the-Invention paragraph, which is not the legal claim scope.

I want to flag this as a known limitation carried over from the earlier sections, not a new contradiction: the earlier "Independent claims — plain-language overview" was explicitly based on the summary language rather than verbatim claims, and that limitation now blocks a rigorous anticipation analysis.

---

Definitive sources to pull the actual citation list

To complete this analysis properly, retrieve these — each will give the face-of-patent "References Cited" (backward) list, which is what § 102 analysis needs:

1. USPTO PatentCenter — Application 15/402,751, document "References Cited" / PTO-892 (Notice of References Cited) and the PTOL-326 (Notice of Allowance) with examiner's reasons for allowance.
   • https://patentcenter.uspto.gov (search application 15/402,751)
2. USPTO Patent Full-Text (PatFT) record for US 9,897,486 — the "(56) References Cited" field.
   • https://ppubs.uspto.gov
3. Google Patents — https://patents.google.com/patent/[US9897486B2](/patent/US9897486B2) — the "Patent Citations" table (backward) and the "Cited By" table (forward). Note the distinction: only the former is § 102/§ 103 prior art.
4. Pre-grant publication — https://patents.google.com/patent/US20170328771A1 — its citations tab, which often lists references carried into the granted patent.
5. Espacenet — https://worldwide.espacenet.com — "Also published as" and cited documents, useful for Korean family members (priority KR 10-2016-0059789).
6. JPO/EPO family search on the KR priority application 10-2016-0059789, since examiner citations in Korea often differ from those in the US and can surface additional art.

Request to you: if you can paste the "(56) References Cited" list (or the PTO-892 / PTOL-326 contents for 15/402,751), I will produce exactly the deliverable you asked for — full citation, publication/filing dates, a short description of each reference, and a claim-by-claim § 102 anticipation assessment — with the statutory ground cited for each.

---

Confidence statement

• High confidence: US 9,897,486 B2 is Samsung's spectroscopic-ellipsometry calibration patent; its bibliographic data (App. 15/402,751; filed 2017-01-10; issued 2018-02-20; priority KR 10-2016-0059789, 2016-05-16) is confirmed from the authoritative text.
• High confidence: The three forward-citation data points above are real and correctly transcribed.
• Low confidence / explicitly unknown: The identity of any prior art cited on '486, and therefore any § 102 claim mapping. I did not find it, and I will not guess at it.

I'll run some targeted searches to try to close the two data gaps that block a rigorous § 103 analysis — the granted claim set and the face-of-patent citation list — before writing the obviousness assessment.

Let me try to pull the granted claims and the face-of-patent citation table directly from patent databases.

The searches confirm the same two gaps flagged earlier (no face-citation list, no verbatim claims), but this round surfaced several genuinely on-point spectroscopy-standardization references. Let me pin down the most important one before writing.

§ 103 Obviousness Analysis — US 9,897,486 B2

Bottom line: This round of research finally broke through one of the two gaps flagged in the prior sections. I located a highly material pre-2016 reference family — the "Metrology hardware specification using a hardware simulator" family (WO 03/083452 A1; US 2003/0187602 A1; US 6,721,691 B2) — whose disclosure maps onto nearly every element of the calibration-method claim family, including the two features the specification presents as its point of novelty. I still do not have the granted claims or the face-of-patent citation list, so what follows is a provisional grounds analysis, not a filing-ready petition. I flag every assumption.

---

1. Scope, limitations, and one new contradiction flag

The two blockers carried forward (unchanged):

New contradiction flag (must be corrected before reliance):

The Prior Art section classified US 9,188,543 B2 (General Electric), KR 10-2014-0056261 A, and US 7,002,693 B1 as forward citations of '486 — i.e., documents that cite '486. That cannot be correct for US 7,002,693 B1, which is a 2006 document and therefore cannot cite a patent that issued 2018-02-20. Either the transposition or the citation direction in that section is wrong. If those three are in fact on '486's face as backward citations, all three fall inside the prior-art window and become § 102/§ 103 candidates. This needs to be resolved from the PTO-892 / the '486 Google Patents page itself.

Minor log note: the task header states the current date as April 26, 2026; the operating context states 2026-09-29. Nothing in this analysis turns on the difference, but I am not silently reconciling them.

---

2. The governing standard

(a) Graham v. John Deere Co., 383 U.S. 1 (1966) — scope and content of the prior art; differences between the prior art and the claims; level of ordinary skill; secondary considerations.

(b) The art is AIA art. '486's effective filing date is 2016-05-16 (KR priority). Pre-AIA § 102(b) does not apply. The applicable gateways are § 102(a)(1) (patented, described in a printed publication, or in public use before the effective filing date) and § 102(a)(2) (effectively filed U.S. applications/patents). Every reference below is a pre-2016 printed publication or patent, so § 102(a)(1) is satisfied on its face for each — subject only to confirming each document's actual publication date.

(c) KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) supplies the motivation rationales: combining prior-art elements according to known methods to yield predictable results; substituting one known element for another to obtain a predictable result; using a known technique to improve similar devices in the same way; and "obvious to try" where there is a finite number of identified, predictable solutions.

(d) A point that matters a great deal here — analogous art. '486's own specification states that "the measuring apparatus is not limited" to SE and includes "all types of apparatuses that are capable of generating a spectrum based on reflected light," expressly listing SE, SR (spectroscopic reflectometry), XPS, XRR, and UPS. That is an express field-of-use statement. It defeats any argument that scatterometry / optical CD metrology art is non-analogous, and it supports a reasonable expectation of success in porting a spectral-calibration technique from one reflected-light-spectrum apparatus to another.

---

3. Provisional element decomposition (from the Summary paragraphs — NOT the claims)

Feature-specific disclosures in the specification that a challenger must reach (these are where any genuine novelty would have to live, and therefore where the § 103 fight will be):

1. The selectivity step (spec's "Step 1": Sm = f(AOI) * f(alignment state of sensor)) — only AOI and sensor/array-detector alignment state are treated as apparatus parameters; light-source wavelength/NA, polarizer angle and analyzer angle are discarded as non-influential.
2. The "ideal spectrum" is a modeled spectrum corresponding to the apparatus parameters in their normal/reference state (AOI = 70°, sensor at reference position), automatically acquired when the sample is determined, with the sample state excluded because "states of samples which are used for error correction… are the same."
3. The inverse-operation calibration: Sc = Sm · (ΔAOI)⁻¹ · (ΔSen-AI)⁻¹ (spec's "Step 3"), built on Sm = Si · (ΔAOI) · (ΔSen-AI) · (ΔRe) (spec's "Step 2").
4. The persistence benefit: one calibration serves all later measurements and process steps; no per-step re-correction.

---

4. Reference universe

Tier A — established in the Prior Art section

Tier B — newly surfaced in this step (all pre-2016; verified as published documents)

Caveat on Tier B: these surfaced via search of third-party pages, not from '486's face or file wrapper. I have not verified (i) whether any of them is cited on '486, (ii) full-text pinpoint cites, or (iii) the assignee of the hardware-simulator family (my recollection is Therma-Wave, Inc. — unverified; confirm on the face of US 6,721,691). The family relationship among WO 03/083452 A1 / US 2003/0187602 A1 / US 6,721,691 B2 is established by identical specification text (Steps 490/500/520 and the ΔV = V − V0, ΔS = J·ΔV, S′ = S + ΔS passages appear in all three), which is solid ground.

---

5. The lead reference: disclosure-to-limitation mapping

This is the single most dangerous reference for the calibration family. The following quotes are from the WO 03/083452 A1 / US 6,721,691 B2 specification as returned in search:

Assessment: The lead reference discloses (i) modelling with ideal hardware parameters, (ii) comparing measured curves against those ideal curves, (iii) solving for a parameter-error vector ΔV whose components are hardware parameters including AOI and NA, (iv) applying a matrix operation on the spectrum as a function of that error vector, and (v) selecting which hardware parameters to include and zeroing the rest. That is a very close read on F1(a)–(d). Notably, it is also normalized to a single tool ("between different pieces of metrology hardware" and "over time for a single piece"), which anticipates the patent's asserted advantage of not re-calibrating per step.

---

6. Proposed grounds of rejection

All grounds are keyed to F1 unless stated. Pinpoint citations to column/line must be added from the full texts before filing — the quotes above are from returns of the specification text, not from a certified copy.

Ground 1 — Anticipation-adjacent / single-reference § 103: WO 03/083452 A1 (or US 6,721,691 B2)

Theory: The reference teaches F1(a), (b), (c), and (d) essentially as recited. If, on claim construction, "calibrating the measuring apparatus" reads on correcting the spectrum produced by the apparatus (which is what the claim's own operative language does), the reference's S′ = S + J·ΔV is that correction, applied to the spectral output of a reflected-light metrology tool, on a parameter-error vector whose components include AOI.

Patent owner's best rebuttals (see §8): (1) direction — the reference perturbs the library/model; the claims calibrate the apparatus; (2) algebra — additive J·ΔV vs. multiplicative (Δ)⁻¹ inverse-matrix; (3) the "ideal spectrum" of the reference is keyed to hardware ideal values, while '486 keys it to the sample (bare wafer) and auto-acquires it.

Ground 2 — WO 03/083452 A1 in view of EP 0 437 249 A3 (Shimadzu)

Adds: error-characterization by systematically varying error factors and correcting via a projection/calibration computation — i.e., the generalized technique of deriving a correction operator from measured error-variation vectors. Why combine: both are spectroscopy-correction methods addressing instrument-induced spectral error; Shimadzu supplies an alternative, well-known linear-algebraic route to the same correction, making the substitution a "use of a known technique to improve similar devices in the same way."

Ground 3 — WO 03/083452 A1 in view of US 6,049,762 / EP 0 560 006 B1 / JP 3547771 B6 (Bio-Rad)

Adds: the express motivation. Bio-Rad states the identical problem in the identical way — instrument profiles "vary from instrument to instrument and even from time to time," and the objective is to make "output spectra… no longer instrument-dependent," implemented as a vector × matrix transformation. Why combine: the Bio-Rad standardization doctrine gives a POSITA both the reason (tool-to-tool matching in a fab fleet) and the mechanism (matrix spectral transformation) to apply the hardware-simulator family's ΔV/J methodology to a measurement spectrum rather than only to a library.

Ground 4 — Any of Grounds 1–3, further in view of Tompkins (1993) — for claim family F2

Adds: the F2 steps that are pure ellipsometric regression — build a model spectrum, iterate sample/device parameters, select the model that matches the measured spectrum, read thickness/shape from the fitted parameters. This is the universal ellipsometric method taught by Tompkins, and it is also the express subject matter of US 2004/0130717 A1 (Drevillon) ("a process for the calibration of this polarimetric system, and a polarimetric measurement process"). Substituting the reference's corrected spectrum for the raw measured spectrum as the regression target is the epitome of a predictable, results-effective variation.

Ground 5 — Ground 4 plus the fab-process step — for claim family F3

Adds: "performing a semiconductor process related to a measurement on the device." The hardware-simulator reference already places the method inside a fabrication system ("integrated into a fabrication system such as a lithography or etch system"). Performing a semiconductor process step between calibration and measurement is conventional inline-metrology practice; the motivation is yield control. Note: F3 is the family most vulnerable to a different problem — a generic process step added to a measurement method rarely supplies patentable weight, but it is also, per the patent's own framing, the claim family whose value is in the manufacturing-flow integration.

---

7. Motivation to combine — enumerated KSR rationales

1. Same problem, same field. Both '486 and Bio-Rad/Hardware-Simulator address spectral measurement error arising from hardware non-ideality and hardware-to-hardware variation, in reflected-light spectroscopy. The lead reference even addresses "wear-induced drifting" of the same parameters (AOI, NA) that '486 selects.
2. Known technique, predictable result. Matrix/vector spectral transformation for instrument matching was a decades-old, well-understood technique (Bio-Rad 1996/2000; Shimadzu 1992; Hardware-Simulator 2003/2004). Applying it yields nothing more than the expected calibrated spectrum.
3. Finite, identified, predictable solutions. Given a hardware-parameter error vector, the options were (i) correct the model, (ii) correct the measurement, (iii) re-derive the calibration curve by orthogonal projection. Three known options, all predictable. KSR "obvious to try."
4. Express teaching to subset parameters. The lead reference itself teaches selecting the subset of hardware parameters to include and zeroing the rest — which is '486's Step 1 selectivity, taught affirmatively rather than merely suggested.
5. Design incentive / market pressure. The specification concedes that fab fleets run "tens to hundreds of SE apparatuses" and "100 or more steps… per product," with SEBC requiring correction at each step. That is a recognized, quantified inefficiency — a classic incentive to simplify to a one-time per-tool calibration.
6. Common ownership / same technical community. All references are the work of optical-metrology and spectroscopy practitioners; the lead reference and co-pending Bao application are from the semiconductor-metrology tool industry.
7. Analogous art is not a defense for the patent owner. By express statement in the specification, "the measuring apparatus is not limited" to SE and includes all apparatuses generating a spectrum from reflected light (SE, SR, XPS, XRR, UPS). Scatterometry/optical-CD art is therefore squarely analogous.

---

8. Where the patent owner has genuine room

I want to be balanced — there are real, non-frivolous non-obviousness arguments:

1. Direction of correction. The lead reference modifies the library (S′ = S + ΔS) so the tool's stored model matches the hardware. '486 modifies the measurement (Sc = Sm·(Δ)⁻¹) so the reported spectrum matches the ideal. If the claims are construed to require correcting the apparatus's output spectrum, this is a real difference. Counter: the reference expressly contemplates calibrating "the library spectra between different pieces of metrology hardware" — the practical effect on reported measurements is identical, and the algebraic inversion of a linear perturbation is a routine design choice.
2. Keying the ideal spectrum to the sample. '486's ideal spectrum is "automatically acquired when the sample is determined" (bare wafer), and the sample parameter is deliberately excluded. The reference's ideal spectrum is keyed to hardware specifications and includes material parameters. Counter: the claim language derived from the Summary requires only that the ideal spectrum "correspond to the apparatus parameters," which the reference satisfies.
3. Apparatus-parameter identity. '486 selects AOI + sensor/array-detector alignment state. Alignment-state error as a distinct apparatus parameter (position/rotation/twist — FIGS. 10A–10C) is not shown in the returns I have. If a dependent claim is limited to alignment-state correction, that limitation may have better support. Counter: the lead reference teaches subsetting any hardware parameters, so selecting alignment state is an obvious application of its own teaching; and US 6,049,762 teaches instrument-profile variation as a first-class error source.
4. Persistence across process steps. '486 emphasizes one calibration surviving all subsequent process steps. Counter: the lead reference expressly calibrates per hardware unit (and over time for a single unit) — structurally the same lifetime behavior.
5. Claim drafting. The 13-month pendency, zero IPRs, and zero litigation (per the earlier sections) are consistent with a claim set that was narrowed during prosecution. If the issued claims carry limitations not visible in the Summary — e.g., a required "substantially consistent" threshold, a specific inverse-matrix formulation, or a specific sequence — the analysis must be re-run against the actual text.

---

9. Secondary considerations (Graham factor 4)

Net: secondary considerations do not, on the present record, rescue the calibration family. The FIG. 15A–15C disclosure is affirmatively helpful to a challenger.

---

10. What would convert this from hypothesis to a grounds document

1. The granted claims of US 9,897,486. Everything in §3 is a proxy. Retrieve from PatentCenter (App. 15/402,751) or the Google Patents claims tab. Without these, no element chart can be finalized and no ground should be filed.
2. The PTO-892 and PTOL-326 for App. 15/402,751. This is the most important single document: it tells you (a) whether the hardware-simulator family was already considered, and (b) the examiner's stated reasons for allowance. If WO 03/083452/US 6,721,691 was not considered, Ground 1 is an "art not previously of record" ground with maximum leverage. If it was considered and overcome, the petition must engage the applicant's distinguishing argument directly.
3. Full certified copies of WO 03/083452 A1, US 6,721,691 B2, US 2003/0187602 A1 with pinpoint column/line and figure cites (FIGS. 4 and 5 of the family appear to be the ΔV/J flow charts). Confirm assignee, priority and publication dates on the face.
4. Confirm the citation direction for US 9,188,543 B2, KR 10-2014-0056261 A, US 7,002,693 B1 (see §1 contradiction).
5. Full text of EP 0 437 249 A3, US 6,049,762, EP 0 560 006 B1, JP 3547771 B6 with pinpoint cites. (I have excerpts only; my verification search on US 6,049,762 was cut off by the step limit.)
6. Claim-construction positions on: "apparatus parameters," "ideal spectrum," "calibrating the measuring apparatus," "substantially consistent," and "an influence on a measurement spectrum."
7. A POSITA declaration fixing the level of ordinary skill (I would propose: an optical-metrology engineer/physicist with an MS and ~3–5 years in semiconductor film/thin-film spectral metrology) and the motivation to combine.

---

11. Confidence statement

• High confidence that the WO 03/083452 A1 / US 2003/0187602 A1 / US 6,721,691 B2 family is real, pre-2016, in the same technical field, and discloses ideal-hardware-parameter modelling, measured-vs-ideal curve comparison, a hardware-parameter error vector ΔV whose example components include AOI and NA, parameter subsetting, and a matrix spectral correction S′ = S + J·ΔV. The quotable text was returned consistently across three separate sources, including the family PDF and the Justia and Google Patents pages for US 6,721,691.
• High confidence that this family is the strongest § 103 reference located so far, and that the patent owner's best defenses are the direction-of-correction and algebraic-form distinctions — both of which are, in my view, vulnerable under KSR.
• Moderate confidence in the Tier B secondary references (dates verified from the search returns; full text not read; assignee of the lead family unverified).
• Low confidence / explicitly unknown: the content of the granted claims; whether any Tier A or Tier B reference is on '486's face; and therefore whether this family was overcome during prosecution. I have not identified any reference as anticipatory under § 102 in the strict sense — US 6,721,691's correction is applied to a model library, and characterizing that as "calibrating the measuring apparatus" is an equivalence argument, not an identity argument.

Recommendation: treat Grounds 1+3 as the primary § 103 theory, with Ground 2 as a reinforcing secondary-teaching ground, and do not file anything until the granted claims and the PTO-892 are in hand. The PTO-892 is the gating document.