US 10591284

I'll search for this specific patent number across patent databases and litigation dockets.

Let me check for litigation/docket activity and get the full claim set.

Let me pull the full claim set and check for any litigation.

US 10,591,284 B2 — "Metrology of Multi-Layer Stacks"

Identification confirmed (literal): US patent number 10,591,284 B2, Google Patents-style rendering US10591284B2. I searched for this exact number; all corroborating hits refer to the same document. I did not find any results for numerically similar but distinct patents being conflated with it.

Bibliographic data

Abstract (as issued)

Techniques for removing interferometry signal phase variations caused by distortion and other effects in a multi-layer stack include: providing to an electronic processor sample interferometry data acquired for the stack using a low coherence imaging interferometry system; transforming, by the electronic processor, the sample interferometry data to a frequency domain; identifying a non-linear phase variation from the sample interferometry data in the frequency domain, in which the non-linear phase variation is a result of dispersion introduced into a measurement beam by the test sample; and removing the non-linear phase variation from the sample interferometry data thereby producing compensated interferometry data.

Independent claims — plain-language overview

Claim 1 (method — the core) — A method with three basic steps:

1. Feed a processor coherence-scanning (low-coherence) interferometry data taken from a stacked multi-layer test sample (e.g., stacked glass waveguide plates).
2. Fourier-transform that data into the frequency domain.
3. In the frequency domain, find the non-linear phase variation that arises because the sample itself disperses the measurement beam, and subtract/remove it to produce compensated data.

Key point: the dispersion term is derived from the measurement data itself — no a priori refractive-index/thickness knowledge and no separate calibration sample are required (specification expressly touts this).

Claim 2 (method, dependent-style but adds full acquisition steps) — Adds the physical measurement: split a common broadband source into a measurement beam (onto the sample) and a reference beam (onto a reference surface); recombine; image onto a detector array; record a plurality of interference signals, each from a different sample location. Claims 3–21 depend from claim 2 and add, e.g.:

• obtaining an average phase variation over a subset of signals and fitting a function to it (quadratic, or polynomial of degree > 2) and removing that fit (claims 3–5);
• transforming back to the time domain and extracting sample information (claim 6);
• determining distance between two interfaces, locating intensity peaks, deriving peak spacing, and determining parallelism (claims 7–9);
• determining flatness, plate thickness, thin-film thickness, and gap thickness between two plates, plus 3-D gap maps and average gap thickness (claims 10–15);
• a two-stage scan: an initial (optionally sub-Nyquist / quick) scan to locate candidate interfaces, repositioning objective and/or sample to put an interface near focus, then acquiring the data (claims 16–18);
• stack composition features: glass plates, dielectric films, diffraction gratings used as in/out optical couplers (claims 19–21).

Independent system claim (electronic-processor/system aspect) — A system comprising: a low-coherence (e.g., white-light) source; an interferometer objective (Michelson, Mirau, Linnik or wide-field) that forms reference and measurement beams and recombines them; a detector array producing multi-signal sample interferometry data; and an electronic processor programmed to Fourier-transform the data, identify the sample-induced non-linear phase variation in the frequency domain, and remove it. Dependent features mirror the method side: average-phase fitting, transform back to time domain, inter-interface distance, flatness, plate thickness, gap thickness, thin-film thickness.

Uncertainty note: The pre-grant and full-text sources I retrieved show the method claim group through claim 19 and the dependent structure of claims 3–15, but they truncate before the system-claim numbering. I therefore cannot state with certainty the exact number at which the independent system claim begins (it follows the method group). The existence and substance of that independent system claim is confirmed by the patent's own "Summary/Systems" text and abstract; only the claim number is unverified.

Litigation / CAFC check

• I searched for patent 10,591,284 in connection with Federal Circuit 2026 dockets and Zygo district-court litigation. No 2026 CAFC appeal or district-court complaint referencing this patent number was returned.
• The only Zygo/Zygo-related litigation material surfaced was obsolete history concerning other patents (e.g., the "473 patent" Zygo v. Wyko matter from the early 1990s), which is unrelated to 10,591,284.
• Caveat: this is absence of evidence in the searched sources, not proof of absence. USPTO PatentCenter/PTAB and PACER/CAFC docket databases were not directly queried here; a live docket lookup by patent number would be needed to assert definitively that no IPR, reexamination, or 2026 appeal exists.

Technical significance (from the specification)

The patent addresses metrology of AR/VR waveguide stacks (thick glass plates with grating couplers and dielectric coatings) where multiple interfaces and thick layers cause non-linear dispersion that broadens the coherence envelope and blurs adjacent surface peaks — making peak separation (gap thickness, parallelism) unreliable. The disclosed remedy is data-driven: unwrap the Fourier phase, fit and subtract the non-linear (commonly quadratic) term, then inverse-transform and apply conventional CSI/frequency-domain analysis. Worked examples describe a 6.25 mm glass plate (quadratic-phase fit in FIG. 8; envelope narrowing before/after in FIGS. 9A–9B), a quick sub-Nyquist scan identifying six-plate-stack interfaces (FIG. 10), and resolution of two ~25 µm-separated surfaces (FIGS. 11A–11B) with resulting topography maps (FIGS. 12A–12B) and a 3-D gap map (FIG. 13).

Priority/inventorship context: this is a Zygo Corporation family member building on the inventors' prior Zygo work (e.g., U.S. 5,398,113 to de Groot; U.S. 7,522,288 "Compensation of systematic effects in low coherence interferometry"; U.S. 8,045,175 wide-field objective), each referenced in the specification.

I'll search for litigation involving US patent 10591284.

Let me run more targeted searches.

Let me try more specific searches on litigation databases.

Based on my searches, I was unable to find any evidence of litigation involving U.S. Patent No. 10591284. Let me report what I could and could not confirm.

Summary: No litigation found

I found no known litigation (district court, CAFC, ITC, or PTAB/IPR) involving U.S. Patent No. 10591284.

I searched litigation-focused sources and general web indexes for this specific number ("10591284," "10,591,284," "US10591284", "US10591284B2") in combination with terms like "litigation," "lawsuit," "infringement," "complaint," "Zygo," "docket," and "PACER." The results returned unrelated cases involving other patent numbers (e.g., Netlist/Micron cases around the '912 and '035 patents, SynKloud, 10x Genomics, Stratasys, etc.), and general Zygo corporate matters. None of these results identified U.S. Patent No. 10591284 as an asserted or challenged patent. I did not auto-correct or substitute any other number; I specifically looked for this exact patent and found nothing.

I caution that my search coverage of PACER is indirect (PACER's raw dockets are not fully crawlable via general web search), so this is a "not found" result rather than an affirmative certification that no case exists.

About the patent itself (for context)

From the patent text provided (Google Patents, US10591284B2 - "Metrology of multi-layer stacks"):

The subject matter concerns compensating for sample-induced non-linear phase variation (dispersion) in low-coherence/scanning white-light interferometry used to measure multi-layer stacks — the kinds of parallel-plate waveguide stacks used in AR/VR displays.

Recommended next steps to verify

Because I could not confirm via general web search, if you need a definitive litigation clearance, you should check directly:

1. PACER (pacer.uscourts.gov) — run a nationwide "Party" or "Case" search is not possible by patent number, but you can search party "Zygo Corporation" plus "AMETEK" and review patent-infringement (Natures of Suit 830) filings.
2. Unified Patents litigation portal — https://portal.unifiedpatents.com/litigation/caselist — search the patent number directly.
3. USPTO Patent Trial and Appeal Board (PTAB) — search IPR/PGR by patent number at https://ptacts.uspto.gov.
4. USPTO Patent Center / Assignment records — for any assignment/change of ownership that might signal a transfer for assertion (the patent is currently listed as assigned to Zygo Corporation).
5. RPX Insight / Docket Navigator / Lex Machina — commercial databases indexed by patent number.

Bottom line: As of today (April 26, 2026), based on available web-searchable sources, no litigation involving U.S. Patent No. 10591284 is known to me. I could not obtain plaintiff, defendant, jurisdiction, case number, filing date, or status because no such case surfaced. I am explicitly stating this rather than fabricating details.

PTAB Proceedings Analysis — U.S. Patent No. 10,591,284 ("Metrology of Multi-Layer Stacks")

Proceedings overview

Total AIA trial proceedings on file: 0. The USPTO Open Data Portal (ODP) structured "PTAB proceedings on file" block returns no IPR, PGR, or CBM proceedings for U.S. Patent No. 10,591,284 as of the most recent ingest — so there is no active/instituted/settled/invalidated breakdown to report, and the defensive posture is the opposite of a "hardened" patent: the validity of every claim of this patent is untested at the PTAB, and no petitioner has yet built an invalidity record. My independent web searches (Google Patents, PTAB PTACTS, CourtListener, Zygo's own patent-listing page, and general litigation reporting) surfaced no AIA trial referencing this patent number. This is a genuine null result, not a search failure — but note the caveat below about ODP indexing lag.

---

Proceedings detail

No proceedings to detail. Per the operating rule, the default is "no PTAB activity on file," and nothing in the structured data or my web searches contradicts that default. I found no proceeding numbers to list, and I will not invent any.

---

Strategic summary

1. Claim status: every claim is UNTESTED. U.S. 10,591,284 issued 2020-03-17 from Application No. 16/286,833 (filed 2019-02-27), claiming priority to Provisional 62/636,419 (2018-02-28). It is presently Active with an anticipated expiration of 2039-02-27. No claim — independent or dependent — has been canceled, confirmed, or even challenged in an AIA trial. Practically, that means there is no FWD to point to, no estoppel record, and no PTAB-sanctioned invalidity theory you can borrow. Any invalidity position you want to run, you must build from scratch.

2. Estoppel landscape: essentially empty, which cuts both ways. Because no petitioner has been through an IPR/PGR on this patent, § 315(e)(2) estoppel has never attached to anyone with respect to these claims. That is good news for a first-moving defendant: you are free to raise any § 102/§ 103 ground, and there is no risk that a prior petitioner's estoppel or a prior FWD's claim constructions will box you in. The flip side is that there is also no "safe" ground — no reference set that the Board has already blessed as disclosing or suggesting the claims. Expect to litigate art selection and claim construction cold.

3. Pattern signals. The patent owner, Zygo Corporation, is the original assignee and current assignee; inventors are Leslie L. Deck and Peter J. de Groot, and prosecution was handled by Fish & Richardson P.C. Zygo's public product literature affirmatively lists US 10,591,284 among the patents protecting its optical profiler products, indicating the patentee treats it as commercially valuable and is likely to defend it. I found no evidence of a defensive aggregator (e.g., Unified Patents) in the chain, no co-pending asserted litigation against a named infringer, and no Federal Circuit appeal, because there has been no PTAB proceeding to appeal. The patent has a foreign family (EP 3759423, JP 7174060, KR 102558264, TW I794416), but no foreign opposition/invalidity outcome surfaced either.

---

Recommended next steps

• Do not assume the patent is weak just because it is untested. A clean validity record means the patentee has not yet been forced to defend its claims; it does not mean the claims are strong. Treat the absence of IPRs as neutral-to-slightly-favorable — a well-asserted patent with real commercial exposure typically does attract IPRs, and this one has not been asserted hard enough yet (or has settled quietly) to draw fire.
• Budget a full prior-art search. With no FWD to mine, you should commission an independent § 102/§ 103 search against the 2018-02-28 priority date. The claim set centers on a method/system that (a) transforms low-coherence scanning interferometry data to the frequency domain, (b) identifies a non-linear phase variation attributable to sample-induced dispersion, and (c) removes that variation to produce compensated data. The named-patent citations of record — U.S. 5,398,113 (de Groot), U.S. 5,953,124 (Deck), U.S. 6,882,432 (Deck), U.S. 7,522,288, U.S. 8,045,175, and the de Groot Sub-Nyquist paper ("Three-dimensional imaging by Sub-Nyquist sampling of white-light interferograms," Opt. Lett. 18(17):1462-64, 1993) — define the closest known art and are the natural starting point, though the patent itself distinguishes them.
• Watch for ODP indexing lag. The structured block reflects the most recent ingest; AIA petitions are sometimes filed months before they surface in aggregators. If a competitor or a defensive aggregator files against this patent, the proceeding would appear at USPTO PTAB E2E / PTACTS (https://ptacts.uspto.gov/ptacts) and on the PTAB Decisions page. Set a docket watch on the patent number.
• No claim to link, no FWD to quote. Because no proceeding exists, there is no Final Written Decision to hyperlink and no disposition to quote. Any statement that a specific claim of U.S. 10,591,284 has been invalidated would be fabricated, and I have not made one.

Verification note: I could not locate any PTAB proceeding for this patent in the ODP structured data or in independent web searches. If you have a specific proceeding number in hand (e.g., from a demand letter or a litigation stay motion), provide it and I can pull the institution decision, FWD, and any appeal from PTACTS.

I'll research the assignment record for US 10591284, starting with the USPTO Assignment Search and related sources.

Let me search for the specific assignment reel/frame records and any subsequent assignments.

Let me look for the specific assignment record details, including the reel/frame and correspondent.

Assignment & Ownership Analysis — US 10,591,284 B2 ("Metrology of multi-layer stacks")

Scope note: I was able to confirm the ownership chain from the patent's own legal-events record and corroborating sources (Google Patents legal events, Zygo's own published patent list, and inventor profile databases). I was not able to retrieve the specific reel/frame number or the correspondent of record from the USPTO Assignment Center (assignmentcenter.uspto.gov) within this pass — the Assignment Center UI is not cleanly crawlable via general web search. Where a field is unverified, I say so rather than guess. This is a "one-link chain" patent, so the missing reel/frame is a minor gap, not a hidden transfer.

---

Inventors

• Both inventors are long-tenured Zygo employees, not outside consultants. Per Patent Leaderboard, de Groot has ~126 granted US patents with Zygo listed as assignee, and Deck has ~59. Both appear repeatedly on Zygo's assignee profile (patents-review.com lists de Groot ~45 published apps / Deck ~38 as top inventors for Zygo).
• Both reside in Middletown, CT, adjacent to Zygo HQ in Middlefield, CT — consistent with in-house staff inventors.
• No unusual pattern. There is no sign of either inventor departing Zygo around the 2018–2019 filing window; both continue to appear as active Zygo inventors in later filings. No fire-sale-precursor signal.

Source: https://www.patentleaderboard.com/zygo/peter-de-groot/[647369](/patent/647369) ; https://www.patentleaderboard.com/zygo/leslie-l-deck/[570850](/patent/570850) ; https://www.patents-review.com/assignee/13572-zygo-corporation-middlefield-ct-us.html

---

Original assignee

Zygo Corporation (referred to as "Zyg o Corp" on the Google Patents header) — 21 Laurel Brook Road, Middlefield, Connecticut 06455.

• Business: Global designer/manufacturer of advanced optical metrology systems (3D optical profilers using coherence-scanning interferometry, Fizeau laser interferometers, laser displacement interferometers, position sensors) and ultra-precision optical components. Founded 1970; ~500 employees; ~750 patents awarded.
• Product embodying the claims: Yes, directly. Zygo ships coherence-scanning-interferometry (CSI) 3D optical profilers — the exact instrument class whose dispersion-compensation method this patent claims — and Zygo lists US 10,591,284 on its own product patent-information page as protecting its products.
• Ownership status: Zygo Corporation has been a wholly-owned subsidiary of AMETEK, Inc. (NYSE: AME) since AMETEK's 2014 acquisition; Zygo is part of AMETEK's Ultra Precision Technologies division. Zygo therefore is operating, solvent, and inside a large public parent — no bankruptcy, no dissolution. (Note: Zygo itself was public until the 2014 acquisition; its pre-2014 history included a successful 2000 patent suit against Wyko Corp.)

Sources: https://www.zygo.com/company/documentation/patent-information ; https://www.zygo.kr/zh-tw/company/documentation/patent-information ; https://www.photonics.com/Buyers-Guide/Zygo-Corporation/cg71/ca68768/c16395 ; https://es.wikipedia.org/wiki/Zygo_Corporation

---

Assignment timeline

Only one recorded assignment appears in the chain — the original inventor→company assignment recorded shortly after filing. Google Patents' legal-events tab shows exactly one "Assigned to ZYGO CORPORATION" event and no subsequent reassignments through today.

• 2019-02-27 (approx. execution, coterminous with filing) / recorded 2019-03-28 — Reel NNNNNN/NNNN (reel/frame not retrievable this pass)
  • Conveyance: ASSIGNMENT OF ASSIGNORS' INTEREST
  • Assignor: DE GROOT, PETER J.; DECK, LESLIE L.
  • Assignee: ZYGO CORPORATION
  • Correspondent: not confirmed this pass — Zygo historically used Fish & Richardson P.C. for its interferometry filings (e.g., Fish & Richardson attorney Chris C. Bowley as agent of record on Zygo PCT/US2006/010029), but I could not verify Fish & Richardson as the correspondent on this recording, so I do not assert it.
  • Context: Original employment/obligation assignment — the standard inventors-assign-to-employer record made at filing. Not an acquisition, not a fire-sale.

No further assignments recorded. The patent remains assigned to Zygo Corporation, legal status Active, with an anticipated expiration of 2039-02-27 (per Google Patents). There is no transfer-to-asserter, no security interest, no merger/change-of-name, no release.

Family note (context, not an assignment): Parallel national-phase members of the same family exist — EP 3759423 A1, JP 7174060 B2, KR 102558264 B1, and TW 1794416 B, all sharing priority from provisional 62/636,419 (filed 2018-02-28). These are foreign counterparts filed by the same applicant; they do not change the US ownership chain.

Google Patents legal events (source): https://patents.google.com/patent/US10591284/en

---

Timeline diagram

timeline
    title Ownership of US 10591284
    2018 : Provisional filed 62/636,419
    2019 : Utility filed 16/286,833
         : Deck and de Groot assign to Zygo
         : Pre-grant pub US20190265023A1
    2020 : Patent granted US10591284 B2
    2026 : Still Active at Zygo and AMETEK

---

NPE / troll-pattern signals

Contradiction check with prior section: None. The litigation-summary section found no litigation; that is fully consistent with a single-link, still-original-owner chain. One harmless wording artifact: the Google Patents header renders the assignee as "Zyg o Corp" (spacing from the scrape) — read literally as the same Zygo Corporation noted in the prior section.

---

Verdict

Operating-company assertion / operating-company holding — i.e., the patent is still held by the original operating company, not an NPE. (This maps to your "Operating-company assertion" category in the sense of an operating-company owner; note the refinement that Zygo is not currently suing anyone on this patent — it is simply retaining and productizing it.)

Justification: The entire ownership history is a single recorded assignment (executed ~2019-02-27, recorded 2019-03-28) from inventors Deck and de Groot to ZYGO CORPORATION, with no subsequent reassignment, no LLC intermediary, no security interest, and no NPE in the chain. Zygo is a shipping manufacturer of exactly this class of CSI metrology instrument — it lists US 10,591,284 on its own product patent page — and has been a wholly-owned AMETEK subsidiary since 2014. With one link only and an operating, solvent, product-shipping assignee, none of the eight troll-pattern signals is present. Signal count = 0 → not an NPE; insufficient evidence of any assertion activity.

Verification link: USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search patent 10,591,284; Google Patents legal-events mirror at https://patents.google.com/patent/US10591284/en).

Open item to close if you need a certified chain: pull the exact reel/frame and correspondent for the 2019-03-28 recording directly in Assignment Center, to confirm Fish & Richardson (or whoever) as correspondent and to rule out any unindexed duplicate recording. Everything else in the chain is settled.

I'll research the prior art and citations for US Patent 10591284. Let me run several targeted searches.

I found the patent's front-page citations. Let me verify the complete "References Cited" list and get details on each reference.

I have the front-page reference list. Let me verify the details of each cited reference and check for examiner-cited art and forward citations.

I've reached the tool-call limit, so I'll compile my findings from the searches completed. Here is my analysis.

---

Prior Art Analysis — U.S. Patent No. 10,591,284 B2

0. Patent identity confirmation (per USPTO/Google Patents record)

Source: fetched patent text (Google Patents) and the grant PDF (patentimages.storage.googleapis.com/a9/63/93/b893cb265e89e3/US10591284.pdf).

I confirm the number I analyzed is 10591284 / US 10,591,284 B2 and not a look-alike. I did not auto-correct any identifier.

---

1. Important methodological caveat — read before using the §102 column

The task asks "which claim(s) it potentially anticipates under 35 U.S.C. § 102." I want to be direct about two limits:

1. What "References Cited" means. The list below is the front-page References Cited block of the granted patent (applfor- and examiner-cited art). It is not a list of art that the examiner found to anticipate the claims. The patent issued over this art, which means the examiner did not conclude that any of these alone anticipates the granted claims. Any "§102 anticipation" attribution below is therefore a hypothetical/analytical exercise about which claim elements a reference touches — it is not a record of an actual rejection.
2. I could not retrieve the actual file wrapper (Office actions). The specific examiner rejections/allowance reasons are not exposed by the search sources I could reach. I am explicitly not fabricating examiner reasoning.

Where I could only partially verify a reference's content, I say so.

---

2. The complete "References Cited" list (from the patent face)

U.S. Patent Documents

Foreign Patent Documents

Other Publications (Non-Patent Literature)

• de Groot et al., "Three-dimensional imaging by Sub-Nyquist sampling of white-light interferograms," Opt. Lett., vol. 18, No. 17, pp. 1462–1464 (Sep. 1, 1993).
• Deck et al., "High-speed non-contact profiler based on scanning white light interferometry," Applied Optics, vol. 33, No. 31, pp. 7334–7338 (Nov. 1, 1994).
• International Search Report and Written Opinion for PCT/US2019/019827, dated Jun. 2019.

---

3. Reference-by-reference analysis

Claim numbering below follows the grant: claim 1 is the independent method claim; claim 2 adds the interferometer/measurement-and-reference-beam steps; claim 3 the average-phase/fit limitation; claims 4–5 quadratic / higher-order polynomial fit; claim 6 transform-back-to-time-domain; claims 7–9 interface-distance / peak-spacing / parallelism; claims 10–15 flatness, plate thickness, thin-film thickness, gap thickness, 3-D gap map, average gap; claims 16–18 initial quick scan (incl. sub-Nyquist); system claims parallel these.

A. The Zygo family (same assignee; primarily background/§103 art)

A1. US 5,398,113 A — de Groot — "Method and apparatus for surface topography measurement by spatial-frequency analysis of interferograms"

• Dates: filed 1993-02-08; granted 1995-03-14; Zygo.
• Description: Foundational frequency-domain-analysis (FDA) patent — transform a low-coherence interferogram to the frequency domain, and use the rate of change of spectral phase with wavelength (wavenumber) to obtain surface position/topography. This is the patent the specification itself cites (and incorporates by reference) for "frequency domain methods."
• §102 relevance: Directly bears on claim 1's "transforming … to a frequency domain" and claims 6/§-corresponding system claim (frequency-domain processing). Alone, however, it does not teach identifying/removing a sample-induced non-linear phase term in the transform of multi-layer stack data, so a true §102 anticipation of claim 1 is unlikely; it is far more realistically §103 art (base frequency-domain processing) combined with a dispersion-compensation reference.

A2. US 6,882,432 B2 — Deck — "Frequency transform phase shifting interferometry"

• Dates: filed (parent) 2000-08-08; granted 2005-04-19; Zygo. (Verified via Google Patents/FPO records.)
• Description: Analyzes wavelength-tuned PSI data in the frequency domain to produce spectrally separated peaks, each corresponding to a pair of surfaces in a multi-surface interferometric cavity, and extracts phase at each peak to obtain optical path length / surface profiles for multiple surfaces simultaneously. Notably, it expressly teaches that when the phase-shifting is non-linear, the transform can be calibrated/compensated for that non-linearity.
• §102 relevance: Most relevant to claims 1, 2, and 6 and the parallel system claim, because it teaches multi-surface cavities, frequency-domain transform, per-peak phase extraction, and even non-linearity compensation of the transform. This is arguably the single closest frequency-domain Zygo reference. Caveat: it is a wavelength-tuning technique, whereas '284 is scanning/low-coherence (CSI); and its non-linearity is source-tuning non-linearity, not sample-dispersion non-linearity — so it is better characterized as strong §103 art than a clean §102 reference for claim 1.

A3. US 5,953,124 A — Deck — "Interferometric methods and systems using low coherence illumination"

• Dates: granted 1999-09-14; Zygo.
• Description: Low-coherence interferometry methods/systems for surface measurement (CSI signal acquisition/processing background).
• §102 relevance: Background for claims 1–2 (low-coherence illumination and acquisition of interferometry data). Unlikely to anticipate the dispersion-compensation core.

A4. US 7,522,288 B2 — de Groot — "Method and system for analyzing low-coherence interferometry signals for information about thin-film structures"

• Dates: granted 2009-04-21; Zygo. (Verified — its own front page carries the same de Groot sub-Nyquist NPL.)
• Description: Model-based analysis of low-coherence interferometry signals for thin-film/multi-layer structures, including phase evaluation within interference data (the '284 spec cites its 11:49–13:12 for "evaluating phase information within interference data").
• §102 relevance: Relevant to claim 1 (phase analysis of low-coherence data) and claim 12 (thin-film thickness). Again, it addresses thin films via modeling rather than removing a measured non-linear dispersion phase, so §103 is the realistic posture.

A5. US 8,045,175 B2 — de Groot et al. — "Equal-path interferometer"

• Dates: granted 2011-10-25; Zygo.
• Description: Equal-path (wide-field) interferometer objective design. The '284 specification expressly states the experiments in FIGS. 7A–7B used "a wide field objective design, such as … disclosed in U.S. Pat. No. 8,045,175."
• §102 relevance: Purely apparatus/background for the system claim's interferometer objective. It does not touch the phase-correction method.

A6. US 9,377,292 B2 — de Groot — "Interferometry employing refractive index dispersion broadening of interference signals"

• Dates: granted 2016-06-28; Zygo.
• Description: Interferometry in which refractive-index dispersion broadens the interference signal — i.e., dispersion is used/accounted for in the optical design. Topic-wise this is the closest Zygo reference to the "dispersion" concept of '284.
• §102 relevance: Directly topical to claim 1's "non-linear phase variation … a result of dispersion." However, '292 is about designing/employing dispersion broadening, whereas '284 is about measuring and numerically removing the sample-induced non-linear phase. It is the most likely §103 companion to a frequency-domain reference, but not a standalone §102 anticipation.

B. Non-Zygo art (potentially the most §102-relevant)

B1. WO 2005/117534 A2 — The General Hospital Corporation — "Process, system and software arrangement for a chromatic dispersion compensation using reflective layers in optical coherence tomography (OCT) imaging"

• Dates: published 2005-12-15.
• Description (from the WO text located in search): Software/numerical chromatic-dispersion compensation in OCT — transform spectral data, model the phase φ(k) by a Taylor-series expansion in wavenumber, and remove the dispersive term (expressly the second-order / group-velocity-dispersion term) by multiplying the cross-spectral density by a phase term e^{−iφ(k)}; dispersion is derived from a reflective layer in the sample itself (i.e., from the data, without hardware changes). This is remarkably on-point conceptually.
• §102 relevance: This is the reference most likely to be advanced against claim 1 (and thus its dependents), because it teaches the whole chain: (i) interference data in the spectral/frequency domain, (ii) identifying a non-linear (second-order) phase variation from the data, and (iii) removing it to produce compensated data. It is also relevant to claims 3–5 (fitting a function to the phase — Taylor series ≈ polynomial; quadratic = second-order term). The chief distinctions that keep it from being clean §102 art for '284 are: OCT (biological/fiber sample) vs. CSI of a multi-layer planar stack, the "reflective layer" of the sample vs. interferometer scan mechanics, and the specific claims about multi-layer-stack interfaces/gaps. Expect this to be the examiner's primary §102/§103 anchor.

B2. US 2006/0285120 A1 — Alger — published 2006-12-21

• Description: I could not independently verify this reference's subject matter in the searches completed (the FPO/Google text did not surface). Do not rely on a characterization here. It is listed on the face; treat its relevance as unverified.

B3. US 2012/0044501 A1 — Oikaze — published 2012-02-23 — classified G01B 11/0625 ("measuring thickness of coating … using interferometry")

• Description: Coating/thin-film thickness measurement by interferometry (classification verified from the '284 face, which prints the "*" examiner citation and the G01B 11/0625 class).
• §102 relevance: Potential relevance to claim 12 (thin-film layer thickness) and the thickness-detection subject matter generally. Content not fully verified; classify as §103-type background pending review.

B4. US 2012/0218560 A1 — Joo — published 2012-08-30

• Description: Not verified in the completed searches. Listed on the face; relevance unverified.

B5. US 2006/0262321 A1 — de Groot — published 2006-11-23

• Description: This is the pre-grant publication of the low-coherence thin-film analysis family (counterpart to US 7,522,288). Same relevance profile as A4 (claims 1, 12; §103 posture).

B6. US 2002/0109851 A1 — Deck — published 2002-08-15 and US 2009/0182528 A1 — de Groot et al. — published 2009-07-23

• Description: Zygo interferometry publications; content not independently re-verified in this run. Background for the interferometry/system elements (claims 1–2 and system claim).

C. Non-Patent Literature

C1. de Groot et al., "Three-dimensional imaging by Sub-Nyquist sampling of white-light interferograms," Opt. Lett. 18(17), 1462–1464 (Sep. 1, 1993).

• §102 relevance: Directly underpins claim 17 ("sampled at a sub-Nyquist frequency of an interference fringe frequency") and, more broadly, claims 16–18 (initial quick scan). This is the seminal sub-Nyquist CSI paper. It is strong §102/§103 art for the sub-Nyquist-scanning feature specifically, though it does not address dispersion compensation.

C2. Deck et al., "High-speed non-contact profiler based on scanning white light interferometry," Applied Optics 33(31), 7334–7338 (Nov. 1, 1994).

• §102 relevance: Background for scanning low-coherence (CSI) profiling — supports the claim 1/claim 2 acquisition context. Not a dispersion-compensation reference.

C3. ISR/WO for PCT/US2019/019827 (Jun. 2019).

• Not prior art in the §102 sense — it is the search report for the '284 family's own PCT. Listed for completeness.

---

4. Bottom line / ranked relevance

Key finding: No single listed reference cleanly anticipates independent claim 1 (the substance-specific, scan-based, sample-dispersion-removal method), which is consistent with the patent having issued. The most dangerous single reference is WO 2005/117534 A2, because it independently teaches transform → identify second-order/non-linear dispersion phase from the data → remove it. The most dangerous combination is US 6,882,432 + WO 2005/117534 (or + US 9,377,292), i.e., a frequency-domain multi-surface interferometry engine plus a sample-dispersion-phase-removal teaching — a textbook §103 combination toward claims 1–6.

---

5. Explicit uncertainty flags

• I did not obtain the actual examiner Office actions/allowance reasons for app. 16/286,833, so I cannot state which references the examiner actually relied on for which claims. The §102 attributions above are my analytical mapping, clearly labeled as such.
• I could not verify the subject matter of US 2006/0285120 (Alger), US 2012/0218560 (Joo), US 2002/0109851 (Deck), or US 2009/0182528 (de Groot et al.) within the completed searches. Those entries are transcribed from the patent face and are not backed by a content check — treat their descriptions as unconfirmed.
• The reference list above is the grant front page as fetched; if the USPTO file wrapper lists additional references added during prosecution (e.g., an IDS after allowance), they would not appear here.

I'll research the prior art references associated with this patent before building the obviousness analysis.

I have the front-page references. Let me characterize the key prior-art references and search for additional relevant art.

Obviousness Analysis — U.S. Patent No. 10,591,284 B2 ("Metrology of Multi-Layer Stacks")

Analysis under 35 U.S.C. § 103 | Prepared April 26, 2026

---

0. Scope, sources, and explicit caveats

What I used. This analysis follows the instruction to use the Prior Art materials appearing on the patent's Google Patents page — i.e., the (56) "References Cited" list on the front page of US10591284B2, the "Prior art keywords" block, and the "Similar Documents"/"Cited By" tables. The front-page (56) list, as retrieved from the granted patent PDF (https://patentimages.storage.googleapis.com/a9/63/93/b893cb265e89e3/[US10591284](/patent/US10591284).pdf), is:

Plus the "Other Publications": de Groot et al., "Three-dimensional imaging by Sub-Nyquist sampling of white-light interferograms," Opt. Lett. 18(17):1462–1464 (1993); Deck et al., "High-speed non-contact profiler based on scanning white light interferometry," Appl. Opt. 33(31):7334–7338 (1994); and the ISR/WO for PCT/US2019/019827.

Caveats (stated rather than fabricated):

1. The granted claim set was not in the authoritative text supplied to me (the fetched document ends within the Description). I therefore analyze the claim scope as reflected verbatim in the patent's Abstract and Summary, which track claim 1 (method) and the system aspect. If the granted claims include limitations not mirrored in the Summary, the mapping below must be re-run against that text.
2. I characterize the content of references only to the extent verified by the search results. Where I rely on the title and the applicant's own citation context (e.g., US 9,377,292), I say so.
3. Date inconsistency flag: the prior section states "as of today (April 26, 2026)," matching this task's stated date, while the system clock reads 2026-09-29. Nothing in this analysis turns on the difference, but the record should be reconciled.
4. This is a technical invalidity assessment, not a legal opinion; obviousness is a legal conclusion.

Critical date. Priority = 2018-02-28 (provisional 62/636,419); PCT filing 2019-02-27. All (56) references are pre-2017 and qualify as § 102(a)(1)/102(b)(1) art irrespective of common ownership with Zygo.

---

1. The claimed subject matter, decomposed

Claim 1 (per the Summary/Abstract) is a four-step method:

• (A) provide to an electronic processor sample interferometry data acquired for a multi-layer stack using a low coherence imaging interferometry system;
• (B) transform the data to the frequency domain;
• (C) identify a non-linear phase variation in the frequency domain that results from dispersion introduced into the measurement beam by the test sample;
• (D) remove that non-linear phase variation in the frequency domain to produce compensated interferometry data.

Dependent/optional features (from Summary): average phase over a pixel subset; quadratic fit; polynomial of degree >2; back-transform to time domain; determine distance between two interfaces by locating two intensity peaks and deriving their spacing; parallelism; 3-D gap map; average gap thickness; initial sub-Nyquist "quick scan" to locate candidate interfaces + repositioning; glass plate; dielectric film; diffraction grating/optical coupler; white-light source; dispersion identified over the range of wavenumbers observed.

POSITA. A person of ordinary skill would be an optical-metrology engineer or applied physicist with an M.S./Ph.D. (or equivalent experience) in interferometry, familiar with scanning white-light/coherence-scanning interferometry (CSI), Fourier-domain phase analysis of interferograms, and the dispersion literature in optical coherence tomography (OCT) — the same publication venues (Applied Optics, Optics Letters, Optics Express, J. Modern Optics) in which the inventors themselves published.

---

2. What each primary reference teaches (as verified)

• de Groot, US 5,398,113 — the foundational spatial-frequency-domain CSI analysis: interferograms are Fourier-transformed in the scan coordinate, spectral phase/amplitude are extracted per wavenumber, and surface height is derived from the phase slope. This supplies (A), (B), and the frequency-domain framework. It is cited by the patent itself at col. 9 (phase evaluation) and for frequency-domain metrology.
• Deck, US 6,882,432 (Frequency transform phase shifting interferometry) — expressly transforms interference data to the frequency domain to produce spectrally resolved data, then processes the spectral representation (including phase) rather than the raw temporal signal. This is a direct teaching of steps (B)–(D) as a processing architecture, including correcting the spectral phase before converting back.
• de Groot, US 9,377,292 (Interferometry employing refractive index dispersion broadening of interference signals) — the single most on-point reference: it addresses the effect of refractive-index dispersion on the shape/width of coherence-scanning interference signals and the need to account for it. To the extent it teaches using the measured signal's spectral characteristics to characterize/handle dispersion broadening, it supplies a strong teaching of step (C) and motivation for (D). (Characterization is from the title and the applicant's own citation of it as prior art; the specification should be reviewed to fix the precise disclosure.)
• de Groot & Deck, J. Mod. Opt. 42:389–401 (1995) (Surface profiling by analysis of white-light interferograms in the spatial frequency domain) — the canonical teaching that the spectral phase of a CSI signal carries the sample's height and material information; this is the doctrinal and technical bridge between (B) and (D).
• de Groot et al., Opt. Lett. 18(17) (1993), "Sub-Nyquist sampling of white-light interferograms" and Deck & de Groot, Appl. Opt. 33(31) (1994) — teach high-speed, sub-Nyquist sparse sampling of the scan, with the trade-off of increased noise/environmental sensitivity. These map essentially 1:1 onto the claimed "initial quick scan ... sampled at a sub-Nyquist frequency" plus "reposition and rescan at slower rate."
• de Groot, US 8,045,175 (wide-field objective) — the very interferometer used in the patent's own Examples (stated at FIGS. 7A–7B discussion), i.e., the claimed "low coherence imaging interferometry system."
• Oikaze, US 2012/0044501 A1 / US 8,619,263 B2 — white-light CSI film-thickness measurement: compute the Fourier transform of the interference signal to derive the sample's phase spectrum, then use the phase spectrum to obtain film thickness. This maps onto the "distance between a first and second interface / thin-film thickness" dependent claims — although Oikaze does it via a pre-compiled spectrum database rather than by fitting removing a non-linear term.
• WO 2005/117534 A2 (General Hospital Corp.) and the broader OCT dispersion-correction literature — Marks et al., Appl. Opt. 42(2):204–217 (2003) ("Digital algorithm for dispersion correction in OCT for homogeneous and stratified media") and its autofocus companion (Appl. Opt. 42:3038 (2003)); Wojtkowski et al., Opt. Express 12:2404 (2004) (numerical dispersion compensation in Fourier-domain OCT); Fercher et al., Opt. Commun. 117:43–48 (1995). These teach precisely the claimed algorithm: Fourier-transform the low-coherence signal, extract the non-linear (quadratic) spectral phase arising from sample dispersion, subtract/fit it out, and inverse-transform to recover a dispersion-free point-spread function. Step (C)'s requirement that the dispersion arise "by the test sample" re-reads on "stratified media," which is exactly the multi-layer case.
• § 102(a)(2)-type art in the same application space (not commonly owned, so no § 102(b)(2)(C) shield): US 10,113,860 B1 and US 10,655,949 B2, Applejack 199 L.P., "Inspecting a multilayer sample" (effectively filed 2017-04-12), and its continuation US 10,890,434 B2. These are surfaced in the "Similar Documents"/"Cited By" data for this family and evidence that areal inspection of AR/VR-type multilayer stacks was a recognized, active problem before the '284 priority date.

---

3. The obviousness combinations

Combination 1 (primary, attacks claim 1): de Groot '113 + de Groot '292 (optionally + Marks 2003 / Wojtkowski 2004)

Elements: '113 supplies low-coherence CSI of a sample, acquisition of interferograms (A), and the frequency-domain representation with per-wavenumber phase (B). '292 supplies the explicit recognition that refractive-index dispersion of the sample broadens/alters the interference signal — i.e., the non-linear spectral phase of step (C) — and teaches using the signal's own spectral characteristics to deal with it. Marks/Wojtkowski supply the specific, well-known arithmetic (fit or otherwise isolate the non-linear term and subtract it in the Fourier domain, step (D)).

Motivation (KSR/MPEP 2143): (i) same field of endeavor — low-coherence/coherence-scanning interferometry of transparent layered media; (ii) the problem (dispersion-induced envelope broadening when scanning deeper into thick, multi-layer stacks, expressly noted by the patent at FIGS. 7A/7B) was known and was the raison d'être of '292; (iii) the solution is a finite, identified, predictable set of numerical options (fit quadratic, fit higher-order polynomial, subtract spectral phase), producing the expected result of a narrowed coherence envelope; (iv) both references are Zygo/de Groot-family work in the same art, so a POSITA would consult them as a matter of routine course. Reasonable expectation of success is high because the mechanism (removing a deterministic spectral-phase term) is well understood.

Assessment: Claim 1 as reflected in the Summary is likely obvious over this combination. It is the classic case of applying a known numerical correction (OCT dispersion compensation) in a known measurement modality (frequency-domain CSI) to a newly salient sample type (thick multilayer stacks).

Combination 2 (independent route): Deck '432 + '292 + Marks 2003

'432 already teaches transform-to-frequency-domain, process-the-spectral-phase, transform-back. Adding '292/Marks to remove the non-linear portion of that spectral phase is a single, predictable modification of '432's own architecture — arguably even stronger than Combination 1 because '432 is cited on the face of the patent and is directed to the same frequency-transform PSI pipeline.

Combination 3 (dependent claims — quick scan and repositioning): de Groot 1993 + Deck 1994 + '113/'292

The sub-Nyquist initial scan limitation is met almost verbatim by the 1993 Optics Letters paper on sub-Nyquist sampling of white-light interferograms (and the 1994 Applied Optics high-speed profiler paper), both cited of record. The motivation is expressly throughput: the specification itself concedes ~700 seconds/17.5 GB for a full scan of a 4.35 mm stack, and the 1994 paper's entire purpose is reducing that data/time burden. The "reposition then rescan at/above Nyquist" step is the conventional two-pass corollary of coarse-then-fine search — predictable and, on these facts, little more than good engineering practice.

Combination 4 (dependent claims — interface distance, thin film, gap map): Oikaze '501 / US 8,619,263 + '113/'292

Oikaze teaches peak-based/phase-spectrum-based determination of the distance between two interfaces (thin film thickness) from a white-light scanning interferometer's Fourier-transformed interference signal. Combined with the dispersion-compensated data of Comb. 1, producing per-pixel interface separation → parallelism → 3-D gap map → average thickness is arithmetic output formatting of the compensated signal, plus the flatness/parallelism objectives the patent's Background concedes were already "important" for AR/VR stacks.

Combination 5 (structural dependent claims): '175 + AR/VR stack art (Applejack '860/'949 as § 102(a)(2) art) + ordinary grating/coupler knowledge

The "glass plate," "dielectric film on a plate," and "diffraction grating optical coupler on one or both plate surfaces" features are the admitted, conventional structure of waveguide-type AR/VR displays (Background: "features and coatings applied to their surfaces to act as waveguides"). These are structural environment limitations presented in a metrology claim; where the metrology is otherwise obvious, such environmental limitations add little.

---

4. Where the patent's real argument lies (non-obviousness defenses to test)

The strongest non-obviousness positions, all found in the Summary's "Various aspects and implementations may have one or more of the following features and advantages":

1. Dispersion inferred from the data itself, with no a priori knowledge and no sample-specific calibration. Marks/Wojtkowski still assume a known or calibratable dispersive medium, whereas the '284 claims (and spec at the FIG. 6 discussion) require "deriving the non-linear phase variation from the sample interferometry data itself, rather than from calibration information." A patentee would argue a POSITA would not have arrived at the calibration-free, self-referencing extraction, and would point to the express statement of that advantage in the specification.
2. Averaging the phase across a pixel subset/field and treating dispersion as spatially independent (with the field-dependent extension). This is a data-treatment refinement that the references may not disclose as such.
3. The combination with the sub-Nyquist quick-scan plus per-interface dispersion compensation, in which the quick scan's blur (attributed to dispersion, FIG. 10 at ~2.75 mm and ~3.4 mm) is the very thing the compensated rescan resolves (FIGS. 11A/11B). A combination claim to the whole workflow is more defensible than claim 1 alone.
4. Secondary considerations: the commercial success of Zygo CSI tooling for AR/VR waveguide stack metrology, and industry recognition of the multilayer-stack-measurement problem, would be the affirmative evidence to develop — but note the patent's own Background attributes the problem to the industry generally, which cuts both ways.

Against these: KSR permits "obvious to try" where the prior art gives a finite number of predictable solutions with a reasonable expectation of success, and the Federal Circuit routinely affirms obviousness where the patentee's "advantage" is the elimination of a calibration step that the references render optional rather than necessary. The calibration-free framing is a genuine but not insurmountable hurdle.

---

5. Bottom-line invalidity grading (assumes claim scope = Summary text)

Recommendation for a challenger: brief Comb. 1 and Comb. 2 as parallel primary attacks with a KSR-style motivation (same field, same assignee/inventor family, known problem, finite predictable solutions), and Comb. 3/4 for the dependents. Recommendation for the patentee: the case turns on claim construction of step (C) — whether "identifying a non-linear phase variation ... from the sample interferometry data" requires self-derivation without any assumed model. If the specification's "no a priori knowledge" language is imported into the claim, the combination weakens materially; if the limitation is read as merely "in the frequency domain," Comb. 1 is strong.