US 5264912

I'll verify this patent number against live sources, including checking for any 2026 CAFC activity.

Let me check for any litigation or court activity tied to this specific patent, and confirm assignee details.

USPTO record — US 5,264,912

I searched for the literal number 5264912 (no similar numbers). Findings below are grounded in the authoritative full text supplied and cross-checked against Google Patents, uspto.report, and FreePatentsOnline.

Assignee note / uncertainty: uspto.report lists the grant as currently assigned to Tencor Instruments, consistent with the "Original Assignee" field. The Google Patents page additionally shows a "Current Assignee" line reading "INTALIMPIANTI OF AMERICA Inc (ITALIMPIANTI) A NY CORP." The only assignment event matching that name on the page is dated 1988-07-26 and involves Blaw Knox Corporation — i.e., it predates this 1992 application and is unrelated to its subject matter. I therefore treat the Italimpianti entry as a database/assignment-record artifact, not a real chain of title for '912, but I flag it because I cannot rule it out authoritatively.

Abstract (verbatim)

"An apparatus used to inspect patterned wafers and other substrates with periodic features for the presence of particles, defects and other aperiodic features in which a spatial filter placed in the Fourier plane is used in combination with either broadband illumination, angularly diverse illumination or both. In contrast to prior devices that direct light from a single monochromatic source through a pinhole aperture stop, embodiments are describes [sic] that illuminate a patterned substrate using (1) a single monochromatic source with a slit-shaped aperture stop for angularly diverse illumination, (2) a single broadband source with a pinhole aperture stop for broadband illumination, (3) a single broadband source with a slit-shaped aperture stop for both broadband and angularly diverse illumination, or (4) multiple sources with an aperture stop for each source for at least angularly diverse illumination. The spatial filters for these illumination systems are characterized by opaque tracks in an otherwise transmissive filter for blocking the elongated bands produced by diffraction from the periodic features on the illuminates [sic] substrate…"

Plain-language overview of the independent claims

Claim 1 — Broadband-source inspection device. An optical inspection tool for patterned wafers where the illumination is a broadband source (e.g., an arc lamp). A beam passes through an aperture stop onto the periodic pattern, which diffracts it into multiple dispersion orders, each forming an elongated band in the Fourier plane. A mostly transparent spatial filter sits in that Fourier plane and carries opaque tracks (each radiating from an imaginary point) that block the pattern-diffracted bands while letting light scattered by aperiodic features (particles, defects) through. Transmitted light is relayed to a two-dimensional imaging sensor. The broadband source is what suppresses speckle; the "track" filter is what handles the resulting spread-out spectrum.

Claim 14 — Multiple-source inspection device. Same architecture, but instead of one broadband source there are several illumination sources aimed at a common region of the substrate, each with its own aperture stop controlling that beam's angular diversity (explicitly contemplated as lasers, same or different wavelengths, spaced at different elevation angles). The transmissive track filter is designed to block the bands from all the sources at once, and again passes aperiodic-feature scatter to a 2D sensor. The diversity here is primarily angular, giving greater speckle reduction than a single illuminator could easily achieve.

Claim 28 — Single monochromatic source with slit aperture. A narrower case: a single monochromatic source, but the key is an elongated slit-shaped aperture stop rather than a pinhole. The slit gives the illumination substantial angular extent in one direction, spreading each diffraction order into an elongated band, which the opaque-track filter is shaped to block. This is the claim directed at laser-based systems where a slit (rather than a pinhole) is used to buy angular diversity and more throughput light while still permitting an effective track filter. Output again goes to a 2D sensor.

Claim 38 — Improvement claim (filter per se, in-system form). A "consisting of"-style improvement to an otherwise conventional patterned-wafer inspection system: the improvement is the mostly-transmissive spatial filter placed in the Fourier plane, having multiple spaced-apart opaque tracks that block the diffracted light bands while transmitting aperiodic-feature scatter, with each track radiating from an imaginary point. This is the broadest, most filter-centric claim; because it is drafted as an improvement over an admitted system, its scope turns on the filter geometry rather than the illumination type.

Dependent-claim themes worth noting: photographic formation of the filter on high-contrast film defocused slightly from the true Fourier plane to oversize the opaque tracks (claims 3, 16, 30); a variable-wavelength/dye or thin-film color blocking filter that blocks only the color actually diffracted to each location (claims 4, 17); slit vs. pinhole aperture stops (6–7, 19–20); normal vs. non-perpendicular illumination via beamsplitter (8–9, 32–33); relay optics to image an internal Fourier plane to an accessible filter location (10, 34); and sensor type — CCD array, vidicon camera, or TDI sensor (11–13, 25–27, 35–37).

CAFC 2026 docket check — no hits

I found no Court of Appeals for the Federal Circuit docket activity in 2026 (or any year) involving US 5,264,912. This is consistent with the record: the patent expired 2012-02-07, roughly fourteen years before today's date, so a 2026 Federal Circuit appeal directly on this patent would be implausible. The CAFC search results returned unrelated matters (e.g., a June 2026 rehearing petition in Spero v. Mercedes-Benz USA, Fed. Cir. No. 25-1306, concerning U.S. Patent 11,208,029 — not this patent).

For completeness on related litigation history: '912's assignee family (Tencor/KLA-Tencor) has extensive inspection-patent litigation, including KLA-Tencor Corp. v. Nanometrics, Inc. (settled/cross-licensed January 2012) and KLA-Tencor Corp. v. Murphy et al. (N.D. Cal. 5:09-cv-01922). I could not verify that '912 itself was asserted in any of those cases, so I make no representation that it was.

Confidence statement

Title, dates, inventors, assignee (original), abstract, status, and the text of claims 1, 14, 28, and 38 are taken directly from authoritative grant text and are high-confidence. Two items carry genuine uncertainty: (1) the anomalous "Italimpianti of America" current-assignee listing, which I believe is an assignment-record artifact but cannot definitively exclude; and (2) the absence of any 2026 CAFC activity, which I state as "none found" rather than a certified negative, since I cannot search every sealed or not-yet-docketed filing.

I'll search for litigation records involving US patent 5264912.

The initial searches returned results for unrelated patents (different '491 patents). Let me refine with more targeted searches.

Let me check specialized litigation databases and the Unified Patents portal directly.

Summary of Findings

No litigation specifically involving U.S. Patent No. 5,264,912 was identified in my searches.

I must be explicit up front: I could not find any district court case, Federal Circuit appeal, ITC investigation, or PTAB proceeding in which U.S. Patent 5,264,912 is listed as an asserted patent-in-suit. I am reporting this as a negative finding rather than fabricating case details.

What the patent is (for context)

• Patent: U.S. 5,264,912 — "Speckle reduction track filter apparatus for optical inspection of patterned substrates"
• Inventors: John L. Vaught, Michael E. Fein, Armand P. Neukermans
• Original assignee: Tencor Instruments (which later became part of KLA-Tencor / KLA Corporation)
• Application: US 07/832,379; filed 1992-02-07; granted 1993-11-23
• Status: Expired – Lifetime; anticipated expiration 2012-02-07
• Family: JP 3458139B2, KR 100246268B1, WO 1993016373A1, TW 243555B

The 20-year term running from the 1992 filing date, with expiration in 2012, defines the window in which infringement litigation could have been (and, if any occurred, would have been) filed.

Searches performed and results

Notably, the searches surfaced other Tencor/KLA-Tencor-related litigation that does not involve this patent, including:

• ADE Corporation v. KLA-Tencor (D. Del., Case No. 1:00-cv-00892) — a competitor dispute, but based on different patents.
• Xitronix Corp. v. KLA-Tencor (W.D. Tex., No. 08-cv-723) and related proceedings — concerning KLA's '441/'486/'260 patents, not 5,264,912.

These are sometimes conflated with the Tencor entity but are not cases about U.S. 5,264,912, so I have not listed them as litigation involving the patent.

Caveats and limitations

1. Database coverage. The public web sources I can query (CourtListener, Unified Patents portal, RPX Insight, Justia, Docket Alarm, courtlistener.com) are weighted toward more recent, high-profile, and post-2010 cases. A short-lived infringement suit filed against 5,264,912 in the 1993–2005 era could exist in PACER or a closed-case archive without an easily retrievable web presence.
2. No auto-correction. Per your instruction, I treated "5264912" literally as U.S. Patent 5,264,912 and discarded all results relating to other patents that share the informal "'491" suffix.
3. KLA-Tencor SEC filings. These increasingly disclose litigation portfolios; my searches did not surface any filing naming 5,264,912 as an asserted patent.

Bottom line

On the record available to me as of April 26, 2026:

• Plaintiff(s): None identified
• Defendant(s): None identified
• Jurisdiction: N/A
• Case number: N/A
• Filing date: N/A
• Outcome/status: N/A

No known litigation involving U.S. Patent 5,264,912 was found. If you need court-grade confirmation, the authoritative next step would be a direct PACER search on the patent number across all districts (including closed cases) and a Lex Machina / Docket Navigator query, which are paywalled and beyond the sources I could reach here. I am flagging this as a genuine gap rather than asserting the absence of litigation as a certainty.

I'll verify the ODP "no proceedings" result against live sources before drafting.

Let me run a couple more targeted checks to be thorough about CBM/PGR and any older proceedings.

AIA Trial Proceedings — Analysis

Source of record: The structured "PTAB proceedings on file" block in my task input (USPTO Open Data Portal, most recent ingest) reports no AIA trial proceedings for US 5,264,912. I ran independent web confirmation against Google Patents, uspto.report, FreePatentsOnline, PTAB decisions portals, and the NYIPLA monthly PTAB institution-decision digests. Nothing contradicts the ODP null result.

Proceedings overview

Total AIA trial proceedings on file: 0. Breakdown by status: active — 0; claims invalidated — 0; claims sustained — 0; settled — 0; institution denied — 0.

The bottom line for a defendant is not "the patent has survived IPRs and is hardened," and it is not "claims 1–5 were canceled." It is a structural point, and it is stronger than either of those: US 5,264,912 was never exposed to the AIA trial regime at all, because it expired on 2012-02-07 — seven months before the first IPR petition could be filed (2012-09-16). There is no FWD to quote, no estoppel to work around, and no PTAB precedent construing its claims. The patent is Expired – Lifetime, so the operative defense against any demand letter citing it is expiration and laches/limitations, not invalidity-through-IPR. Any assertion today would have to rest entirely on pre-expiration conduct within the § 286 six-year lookback — which itself ran out on 2018-02-07.

Proceedings

None. Per the constraint against fabricating proceeding numbers, I am not listing any. What follows is the structural explanation of why the docket is empty, since the absence is itself the signal.

No proceeding — the AIA regime post-dates the patent's enforceable life

• Type: N/A — none of IPR, PGR, or CBM was ever available-and-used against this patent.
• Why no IPR: The AIA's post-grant provisions (IPR, PGR, CBM) apply to petitions filed on or after 2012-09-16. US 5,264,912's 20-year term from the 1992-02-07 filing date ran out on 2012-02-07. In theory the Board can institute IPR on an already-expired patent (the remedy is limited to past damages), and a narrow defensive window existed from 2012-09-16 until the six-year damages lookback closed on 2018-02-07. No petition appears to have been filed in that window. (The window analysis is my inference from § 286; the absence of any petition is confirmed by ODP and by search.)
• Why no PGR: Post-grant review is available only for patents with an effective filing date on or after 2013-03-16. This patent's priority date is 1992-02-07. PGR was legally unavailable from day one.
• Why no CBM: Covered Business Method review ran 2012-09-16 to 2020-09-16 and required a patent claiming a method or apparatus for performing data processing or other operations used in the practice, administration, or management of a financial product or service. US 5,264,912 claims optical wafer-inspection hardware (broadband/slit-aperture illumination, Fourier-plane track filter, 2D CCD/vidicon/TDI sensor). It is not a financial-services patent and was not CBM-eligible. No CBM could have been filed.
• Judge panel: N/A.
• Petition grounds: N/A — no petition.
• Institution decision: N/A.
• Final Written Decision: N/A. There is no FWD addressing claims 1–39, so I make no representation about any claim's patentability, cancellation, or survival. Claims 1, 14, 28, and 38 are all UNTESTED in any AIA forum.
• Settlement / termination: N/A.
• Appeal: N/A — consistent with the earlier CAFC check finding no 2026 (or other) Federal Circuit appeal on this patent. No FWD means no appealable Board decision (pre-AIA § 145-style or AIA § 319 review is unavailable).
• Defensive value: The absence of PTAB history cuts against a would-be petitioner today. The Board has recently shown significant willingness to discretionarily deny institution on patents that have been in force six or more years, on "settled expectations" grounds (a Director-level doctrine applied in 64 of 71 such decisions in mid-2025 per the analyses surfaced in search). A patent that expired in 2012 is the extreme case of that doctrine — an IPR petition against it now would be fighting both the expiration and the discretionary-denial posture. But that is academic: the patent cannot be asserted for any conduct after 2012-02-07, so no IPR is needed.

Strategic summary

Claim status. There is no cancellation record, because there is no PTAB record. Every claim of US 5,264,912 — independent claims 1, 14, 28, 38 and all dependents — is UNTESTED rather than sustained or canceled. A defendant should not represent to a court or an adversary that any claim has been held invalid. The correct and stronger statement is that the patent's entire term has run: it is Expired – Lifetime, with anticipated expiration 2012-02-07. For the family, JP 3458139B2, KR 100246268B1, WO 1993016373A1, and TW 243555B are likewise all marked not-active/expired in the Google Patents family table, so there is no live foreign counterpart to worry about either.

Estoppel landscape. § 315(e)(2) estoppel is irrelevant here because no one ever filed an AIA petition. No petitioner, real party in interest, or privy is barred from raising any prior-art ground. Every ground that a defendant might have wanted to run — including the references cited on the face of the patent (US 3,658,420 Axelrod; US 3,729,252 Nelson; US 3,790,280 Heinz; US 3,972,616 Minami; US 3,981,562 Anthon; US 4,806,774 Lin; US 4,895,446 Maldari; US 5,172,000 Scheff; US 5,177,559 Batchelder) plus the two non-patent citations (George & Jain, Applied Optics vol. 6, p. 1202 (June 1973); Thomas, Applied Optics vol. 7, no. 3, p. 517 (March 1968)) — remains fully available in district court under §§ 102/103/112. Notably, US 5,172,000 (Scheff et al., Insystems) and US 5,177,559 (Batchelder et al., IBM) issued after the '912 filing but before its grant and are already of-record; they, and sibling Tencor application Ser. No. 832,379's own family (e.g., the adaptive-light-valve filter of US 5,276,498, Galbraith et al.), are the natural § 102/§ 103 starting points if anyone ever did assert the patent.

Pattern signals. None of the usual PTAB signals exist: no serial petitioner, no defensive aggregator (Unified Patents, RPX) in the chain, no patent-owner appeal activity, no joinder practice. The reason is timing, not merit — this patent's day in an AIA forum never came, because it predates the forum. The earlier litigation summary similarly found no case in which '912 was an asserted patent, which is consistent with the expiration explanation.

Flagged inconsistency with prior sections. The previously generated litigation summary dates its findings "as of April 26, 2026." Today's date is 2026-09-30. The five-month gap does not change any conclusion here (no PTAB activity has a reason to have appeared in that window for a patent that expired in 2012), but the two sections should be read as having different as-of dates.

Recommended next steps

1. If you are a defendant and a demand letter cites US 5,264,912: do not reach for an IPR. There is no PTAB history to leverage and no claim has been invalidated. Your dispositive point is the front page: Expired – Lifetime, anticipated expiration 2012-02-07. Enforceable only for pre-2012-02-07 conduct, and even that is time-barred after 2018-02-07 under § 286. I have no FWD to link because none exists.
2. If you are a defendant facing live assertion anyway (e.g., past-damages theory): all §§ 102/103/112 grounds are open — no § 315(e)(2) estoppel attaches to anyone. Start with the nine of-record references listed above and the Thomas 1968 and George & Jain 1973 articles, both of which the applicant itself conceded as germane to angularly/spectrally diverse spatial filtering.
3. If you are considering a defensive IPR filing anyway (for a co-pending suit on past conduct): weigh the practical reality that the Board is currently inclined to discretionarily deny institution on long-quiet patents under "settled expectations," and that three named precedential Director Review decisions from February–March 2026 (Revvo Technologies, Generac Power Systems, and the YMTC line) have vacated grants and denied institution on discretionary grounds. An IPR against a patent expired since 2012 fits that doctrine squarely.
4. If you need court-grade confirmation: the authoritative query is the PTAB E2E / PTACTS search at https://ptacts.uspto.gov on patent number 5,264,912, plus an ODP API pull. ODP returned null as of the current ingest; I did not find a PTAB E2E case page for this patent to link.

Caveat on the negative finding. I state this as "no proceedings found," not as a certified negative. ODP coverage of any pre-2018 CBM activity and of any ex parte reexamination (a pre-AIA proceeding outside the AIA trial block and outside the scope of this task) is outside what I can fully verify from public web sources. Nothing on the '912 front page indicates a reexamination certificate, and no reexam was surfaced in search, but I flag that as unverified rather than confirmed.

I'll research the assignment record for US 5,264,912, starting with the authoritative text I already have and then trying to pull reel/frame and correspondent data.

Let me try to pull the specific assignment records and reel/frame numbers.

Let me try to access the USPTO Assignment Center and PatentCenter directly for this patent.

I hit the search limit before I could pull reel/frame-level records from the Assignment Center. Below I reconstruct the chain from the authoritative grant text (which surfaces the USPTO-recorded "reassignment" events) plus uspto.report, and I flag explicitly what I could and could not verify.

Assignment record — US 5,264,912

Scope note / source honesty. The authoritative Google Patents text for this grant displays four USPTO "reassignment" events. It does not display reel/frame numbers or correspondents for any of them, and my attempts to pull those fields from the USPTO Assignment Center index (assignment.uspto.gov / assignmentcenter.uspto.gov) and uspto.report did not return reel/frame data before I exhausted my search budget. I therefore report the events I can ground, and I mark reel/frame and correspondent as not retrieved rather than guessing. This is a real gap in my output, not a finding.

Inventors

Unusual-pattern check. No departures are evident. All three inventors assigned to Tencor within ~3 weeks of filing (1992-02-07 filing; 1992-02-28 recordings), which is the ordinary "inventors assign to employer contemporaneously with filing" pattern. There is no evidence of inventors leaving the assignee within 12 months of filing, and no fire-sale precursor. Note that Neukermans is a prolific Silicon Valley inventor who later founded Xros (optical switching) — but that is a post-Tencor career fact and is not a signal about this patent's chain.

Original assignee

Tencor Instruments Inc. (assignee as listed at grant; also appears as current assignee on uspto.report).

• Primary line of business: semiconductor process-control and wafer-inspection equipment (surface inspection, film-thickness measurement, particle detection). Tencor was a direct competitor in the metrology/inspection space that this patent addresses.
• Did they ship a product embodying the claims? Almost certainly yes in the general sense — Tencor was an operating instrument maker and this patent is directed to its core product line (patterned-wafer optical inspection with a Fourier-plane spatial filter). I could not independently tie this specific patent to a named commercial model number, so I state it as strong inference from the corporate line of business, not a verified product mapping.
• Current status (the original assignee): Tencor Instruments merged with KLA Instruments in 1997 to form KLA-Tencor, now KLA Corporation. So the original assignee no longer exists as a separate entity; it was absorbed into an operating public company — not dissolved, not bankrupt.

Assignment timeline

The authoritative grant text shows these recorded events (chronological):

• 1988-07-26 (executed) / recorded 1988-07-26 — Reel not retrieved

  • Conveyance: Assignment
  • Assignor: Blaw Knox Corporation
  • Assignee: INTALIMPIANTI OF AMERICA INCORPORATED (ITALIMPIANTI), A NY CORP
  • Correspondent: not retrieved
  • Context: Artifact / unrelated chain. This event predates this patent's 1992 filing by ~3.5 years and involves Blaw Knox (heavy-industry equipment), not any party to the '912 chain. It is a display artifact on the Google Patents page (the page labels it a "Current Assignee" line, which is likewise anomalous). This is not part of the ownership history of US 5,264,912 — consistent with the flag already raised in the earlier Patent-summary section above. Do not treat it as a transfer.

  (Cross-reference / contradiction flag: The earlier "Patent summary" section flagged this same Italimpianti entry as a probable assignment-record artifact. My independent review agrees, and I reinforce it: the 1988 date is dispositive.)

• 1992-02-28 (executed) / recorded 1992-02-28 — Reel not retrieved

  • Conveyance: Assignment ("ASSIGNMENT OF ASSIGNORS INTEREST")
  • Assignor: Michael E. Fein
  • Assignee: Tencor Instruments
  • Correspondent: not retrieved (no attorney/firm surfaced in accessible sources)
  • Context: Normal inventor-to-employer assignment. Fein's individual interest assigned to Tencor ~3 weeks after filing.

• 1992-02-28 (executed) / recorded 1992-02-28 — Reel not retrieved

  • Conveyance: Assignment ("ASSIGNMENT OF ASSIGNORS INTEREST")
  • Assignor: Armand P. Neukermans and John L. Vaught
  • Assignee: Tencor Instruments
  • Correspondent: not retrieved
  • Context: Normal inventor-to-employer assignment. Joint assignment of the two remaining inventors' interests to Tencor, same date as Fein's.

No post-issuance assignments appear in the record. No assignment, security agreement, merger, change-of-name, license, or release following the 1993-11-23 grant is surfaced. This is itself a finding: the chain terminates at the original operating assignee.

If the Assignment Center returns no additional records: that conclusion stands — the only recorded transfers are the two 1992-02-28 inventor-to-employer assignments, and ownership remained with Tencor (→ KLA-Tencor → KLA Corporation) throughout the enforceable life of the patent.

Timeline diagram

timeline
    title Ownership of US 5264912
    1992 : Filed by Tencor Instruments
         : Inventors assign to Tencor
    1993 : Patent granted
    1997 : Tencor merges into KLA Tencor
    2012 : Patent expires

NPE / troll-pattern signals

1. Shell-entity transfer — Not present. No transfer to any "IP / Patents / Licensing / Holdings / Ventures" entity is recorded. The only assignee in the chain is Tencor Instruments, an operating instrument manufacturer.

2. Known asserter in the chain — Not present. No assignee matches the public NPE lists (Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Conversant/Mosaid, etc.). Tencor/KLA are operating companies, not asserters in the NPE sense.

3. Repeat correspondent across the chain — Unclear / not retrievable. I could not obtain the correspondent of record for either 1992-02-28 assignment, so I cannot assess recurrence. No correspondent data surfaced on any indexed source I could reach. Reporting this as a gap.

4. Cascading transfers — Not present. Only two assignments, both dated the same day, both inventor→original-employer. No chained LLCs, no <24-month cascades.

5. Pre-litigation transfer — Not present / inapplicable. No infringement suit naming this patent was identified (per the Litigation summary section above), so there is no pre-suit transfer window to evaluate. The patent expired 2012-02-07.

6. Bankruptcy fire-sale — Not present. Tencor was not a debtor; its 1997 event was a merger with KLA Instruments, not a Chapter 7/11 disposition. KLA remains an operating public company.

7. Privateering — Not present. No transfer from Tencor/KLA to any NPE-style entity is recorded, and no evidence of a proxy-assertion arrangement surfaced.

8. Defensive aggregator (anti-NPE) — Not present. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. It terminates at the original operating assignee (KLA lineage).

Verdict

Insufficient data (no post-issuance assignments; only the original inventor-to-employer assignments).

Justification: the recorded chain consists solely of two 1992-02-28 inventor-to-Tencor Instruments assignments (Fein, and Neukermans + Vaught) executed ~3 weeks after the 1992-02-07 filing — i.e., ordinary employment assignments, with no subsequent transfer of any kind through expiration. The lone anomalous display entry (Blaw Knox → Italimpianti, 1988-07-26) predates this application and is not part of the '912 chain. Absent any post-issuance assignment and absent any identified assertion, no NPE signal fires; the correct call is the defined "only the original assignment" branch.

Verification link: USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search patent number 5264912); mirror index at https://assignment.uspto.gov/patent/index.html.

Open items I could not close: (1) reel/frame numbers for the two 1992-02-28 recordings; (2) the correspondent of record for each; (3) any non-web-indexed post-issuance record. A direct Assignment Center query on patent number 5264912 would resolve all three.

I'll research each cited reference against the '912 patent. Let me run targeted searches in parallel.

Now let me research the remaining cited references, including the two post-filing-date Insystems/IBM patents.

Prior Art Analysis — US 5,264,912

Building on the USPTO record already established (grant no. 5,264,912, app. 07/832,379, filed & priority 1992‑02‑07, issued 1993‑11‑23, Tencor Instruments), this section addresses only the prior‑art question: the nine U.S. patents cited on the face of '912, the two non‑patent citations, and their §102 exposure.

1. Legal framework and the critical date

• '912 has no earlier priority claim — its effective filing date is 1992‑02‑07. Because it was filed before 2013‑03‑16, pre‑AIA 35 U.S.C. §102 governs.
• Two dates matter for each reference:
  • §102(a)/(b) ("printed publication" / patent) — the reference's publication (issue) date must precede 1992‑02‑07. Any reference issued before 1991‑02‑07 is a §102(b) statutory bar (published more than one year before filing).
  • §102(e) — a U.S. patent granted on an application filed before 1992‑02‑07, "by another." The reference's issue date can post‑date 1992‑02‑07 and it still counts.
• Key finding up front: Of the nine cited U.S. patents, seven are §102(a)/(b) art and two are §102(e)‑only art (they issued after '912 was filed but were filed before). This distinction matters because those two cannot be used as §102(a)/(b) references at all.

2. The nine cited U.S. patents

2.1 US 3,658,420 — "Photomask Inspection by Spatial Filtering"

• Citation: U.S. 3,658,420; filed 1969‑12‑10; issued 1972‑04‑25; assignee Bell Telephone Laboratories, Inc. (§102(b)).
• Disclosure: He‑Ne laser 10 → pinhole beam‑expander/collimator 11 → rectangular slit aperture 12 (6 × 50 mil) → photomask on a step‑and‑repeat stage → transform lens 16 → a "form factor" spatial filter 17 that is an opaque CROSS stop in the transform plane → reconstruction/magnifier lenses → beamsplitter to a TV camera 24 / monitor 25 (and a PMT). The filter is mounted to be moved into and out of the beam. It departs from Watkins' dot‑array filter and uses a shape (cross) matched to the predominant rectangular mask geometry; scattered light is reduced by the slit and by tilting the mask.
• Element overlap with '912: slit‑shaped aperture stop (claim 6); transmissive Fourier‑plane filter carrying opaque blocking geometry (claims 1/38); a 2D TV camera (claims 11–13 family); monochromatic laser source.
• Missing for independent claims: it is a coherent single‑line laser, not the "broadband source" of claim 1; there is only one source (no "plurality of sources … each having an aperture stop," claim 14); the filter is one connected cross, not "a plurality of spaced apart opaque tracks," and there is no spectral‑dispersion band formation (claim 1/28).
• Potential §102 target: best positioned against claim 28 (single monochromatic source + slit aperture + Fourier‑plane blocking + 2D sensor) and against the filter‑centric claim 38, subject to two genuine gaps: (i) the preamble's "patterned semiconductor wafers" vs. a transmissive photomask, and (ii) "plurality of spaced apart opaque tracks" vs. a single cross. Realistically this is strong §103 art, not a clean §102 anticipation.

2.2 US 3,729,252 — "Optical Spatial Filtering With Multiple Light Sources"

• Citation: U.S. 3,729,252; filed 1970‑06‑05; issued 1973‑04‑24; assignee Eastman Kodak Company (§102(b)).
• Disclosure: An optical system that projects an image‑bearing photographic transparency using a plurality of light beams (point or semi‑point sources) displaced from the optic axis, with a spatial filter in the common (Fourier) image plane having multiple light‑attenuating areas, each in registry with a respective zero‑order beam, to modify image contrast. It expressly notes that increasing pinhole size and/or wavelength bandwidth reduces coherence and increases throughput at the cost of spectral discrimination.
• Element overlap with '912: the "plurality of sources, each with an aperture" concept underlying claim 14; the attenuation/blocking‑filter‑in‑the‑transform‑plane idea (claim 38); the general acknowledgment that source diversity trades against spatial discrimination.
• Missing for independent claims: it is not an inspection apparatus — no patterned wafer, no aperiodic defects/particles, no defect imaging, no 2D electronic sensor; the filter attenuates zero‑order for contrast, it does not block a set of elongated dispersion bands; no broadband arc source.
• Potential §102 target: partial support only for claim 14 (multi‑source/angular‑diversity element). It cannot anticipate any independent claim because the claimed wafer‑inspection subject matter is absent.

2.3 US 3,790,280 — "Spatial Filtering System Utilizing Compensating Elements"

• Citation: U.S. 3,790,280; filed 1972‑05‑03; issued 1974‑02‑05; assignee Western Electric Co., Inc.; inventors R. A. Heinz et al. (§102(b)).
• Disclosure: Laser‑illuminated photomask inspection with a spatial filter comprising a matrix‑like array of opaque regions on a transparent field placed at the back focal plane. The opaque regions block the DC and low‑spatial‑frequency periodic feature information (the "interference‑function" dots); compensating elements block the corresponding higher spatial frequencies (feature‑edge information) to raise S/N in an automated TV‑camera/counter system. It discusses the practical failure of planar filters at their outer regions (defocus/aberration) and solves it with compensating elements.
• Element overlap with '912: the filter‑centric claim 38 ("mainly transmissive spatial filter … a plurality of spaced apart opaque [regions] … blocking pattern light but transmitting non‑periodic light"), and the general architecture of claims 1/28.
• Missing for independent claims: filter geometry is a matrix of dots/lines matched to the interference function, not "tracks each radiating from an imaginary point"; a laser source, not broadband; no angular diversity; no elongated bands.
• Potential §102 target: the single most relevant reference for claim 38 (and claim 39), again with the "radiating tracks" limitation as the gap. Note a nuance for the dependents: '280 treats transform‑surface defocus as a problem to be corrected, so it actually runs against the defocus‑based filter‑forming strategy of claims 3/16/30 rather than anticipating it.

2.4 US 3,972,616 — "Apparatus for Detecting the Defects of the Mask Pattern Using Spatial Filtering"

• Citation: U.S. 3,972,616; filed 1974‑09‑09; issued 1976‑08‑03; assignee Tokyo Shibaura Electric Co., Ltd. (§102(b)). (This is the reference '912's own BACKGROUND expressly discusses and distinguishes.)
• Disclosure: Two sources on the same optical axis — a coherent He‑Ne laser and an incoherent incandescent lamp behind a color filter — combined via a half mirror onto a photomask; a transform lens forms the Fourier‑transform pattern at the filter; the spatial filter 16 is wavelength‑selective and shaped to the FT of the proper mask pattern, blocking the coherent pattern light while letting the (differently colored) incoherent light pass; the screen shows the coherent defect image superimposed on the incoherent pattern image (defects red, pattern green).
• Element overlap with '912: two sources combined on one axis (claim 14); a wavelength‑selective blocking filter (claims 4/17).
• Missing — and teaching away: '616's filter is deliberately designed so the broadband/incoherent light passes to preserve whole‑pattern information. That is the opposite of claim 1, where the broadband light is the very light whose pattern‑diffracted bands are blocked. '616 therefore teaches away from '912's central combination and is a poor §102 reference against claim 1 — but a strong §103 reference on the multi‑source and wavelength‑selective‑filter features.
• Potential §102 target: only claim 14 (multiple sources + Fourier filter) and, as a feature, claims 4/17. No full anticipation.

2.5 US 3,981,562 — "Spatial Filtering for Error Detection" ⚑ most under‑appreciated citation

• Citation: U.S. 3,981,562; filed 1974‑09‑09; issued 1976‑09‑21; assignee Optical Coating Laboratory, Inc.; inventor Erik W. Anthon (§102(b)).
• Disclosure: A Fourier‑transform imaging microscope for repetitive striped parts (striped dichroic color‑TV filters). It deliberately uses non‑coherent monochromatic light of 10–20 nm bandwidth from a variable‑wavelength monochromator (filter/grating/prism) instead of laser light; the monochromator slit is imaged through the striped part to form a Fourier‑transform image consisting of a multiple diffraction image of the slit; slit means suppress the regions of maximum diffraction energy (the stripe‑pattern orders) and view regions of minimum energy, enhancing defects; the wavelength can be varied; a low‑power microscope views the result.
• Element overlap with '912: non‑laser, wavelength‑variable illumination with a slit and Fourier‑plane blocking of repetitive‑pattern energy maxima — conceptually the "diversity + filter" core; the variable‑wavelength aspect of claims 4/17; the slit of claims 6/28.
• Missing for independent claims: the source is only quasi‑monochromatic (10–20 nm), not the "broadband source" of claim 1, and it is not a multi‑source system (claim 14); the filter is slit‑based rather than a plurality of tracks radiating from a point; the instrument is a viewing microscope, not a wafer defect scanner with a 2D electronic array.
• Potential §102 target: most credibly against claims 4/17 (variable‑wavelength blocking) and as the best §103 complement for the "spectral/angular diversity + filter" rationale of claims 1/28.

2.6 US 4,806,774 — "Inspection System for an Array of Microcircuit Dies Having Redundant Circuit Patterns"

• Citation: U.S. 4,806,774; filed 1987‑06‑08; issued 1989‑02‑21; assignee Insystems, Inc. (§102(b)). (Expressly discussed in '912's BACKGROUND.)
• Disclosure (per '912's own characterization plus general knowledge — see verification flag): an inspection system with a Fourier‑transform lens and an inverse Fourier‑transform lens on an optic axis, producing from an illuminated area of a patterned wafer a spatial‑frequency spectrum whose components are selectively filtered to produce a defect image; the filtered image strikes a 2D photodetector array that detects defects in the illuminated on‑axis die, exploiting redundancy in the die patterns.
• Element overlap with '912: the entire claimed optical train of claims 1/14/28 except the illumination/filter innovations — Fourier lens, inverse‑transform lens, 2D sensor ("means for delivering light … to said imaging sensor"), and the preamble of claim 38 ("inspection system … diffracting light … in a Fourier plane from patterned semiconductor wafers … periodic features … defects and particles"). Sensor dependents 11/25/35 (2D CCD array) read directly on '774's 2D photodetector array.
• Missing for independent claims: illumination is monochromatic/collimated (the '912 BACKGROUND itself states the prior art was thought to require "a monochromatic and collimated illumination source, usually a laser"); the filter is a spot/pattern filter, not a plurality of opaque tracks matched to elongated bands; no broadband source (claim 1), no multi‑source/aperture‑stop array (claim 14), no slit (claim 28).
• Potential §102 target: no full anticipation, but '774 is the best §103 base — it supplies everything but the claimed illumination diversity and track‑filter geometry. ⚑ Full text not retrieved in this session; element mapping must be verified.

2.7 US 4,895,446 — "Particle Detection Method and Apparatus"

• Citation: U.S. 4,895,446; filed 1986‑10‑23; issued 1990‑01‑23; assignee Inspex Incorporated; inventors Maldari et al. (§102(b)). (Expressly discussed in '912's BACKGROUND, FIG. 3.)
• Disclosure: a mask in the Fourier plane filters light diffracted from a patterned wafer; a lens forms the Fourier transform of the surface on the mask, the mask containing a pattern corresponding to the Fourier transform of the patterned surface, so that all light except that scattered from particles is blocked; a lens images the passed light onto a camera.
• Element overlap with '912: the general architecture of claims 1/14/28/38 — transmissive Fourier‑plane filter blocking periodic‑pattern light while transmitting particle scatter, imaged onto a camera.
• Missing for independent claims: the mask is a pattern replicating the FT (spots), not "a plurality of spaced apart opaque tracks … each radiating from an imaginary point"; no broadband/angularly diverse illumination; no band formation/blocking.
• Potential §102 target: claims 1/38 on architecture only; the illumination and track limitations prevent full anticipation. ⚑ Full text not retrieved; verify.

2.8 US 5,172,000 — ⚑ §102(e) reference — highest‑priority to obtain

• Citation: U.S. 5,172,000, "Spatial Filter for Optically Based Defect Inspection System"; filed 1990‑11‑02; issued 1992‑12‑15; assignee Insystems, Inc.
• Posture: Not §102(a)/(b) art (it issued ~10 months after '912 was filed). It is a §102(e) candidate because its application was filed 1990‑11‑02, before '912's 1992‑02‑07 filing, "by another." Its presence on the '912 face indicates the examiner had it before him.
• Disclosure: by title and assignee (same company as '774) it is directed to a spatial filter for optically based defect inspection — squarely in the '912 field. I could not retrieve its specification in this session, so I will not describe its disclosure or assert any element mapping.
• Potential §102(e) target: must be run element‑by‑element against claims 1, 14, 28, 38 and the track‑geometry dependents (3, 5, 16, 18, 29, 30, 31, 39). If it discloses a Fourier‑plane filter with strips/bands matched to diffracted light, claim 38 (and possibly 39) could be exposed. Flagged as unverified.

2.9 US 5,177,559 — ⚑ §102(e) reference

• Citation: U.S. 5,177,559, "Dark Field Imaging Defect Inspection System for Repetitive Pattern Integrated Circuits"; filed 1991‑05‑17; issued 1993‑01‑05; assignee International Business Machines Corporation.
• Posture: same as '000 — §102(e) only (issued after '912's filing, filed before, by another).
• Disclosure: the title indicates dark‑field inspection of repetitive‑pattern ICs — i.e., blocking repetitive‑pattern diffraction and imaging scattered/defect light in a Fourier/dark‑field plane. Not retrieved this session.
• Potential §102(e) target: claims 1/38 (Fourier‑plane filtering of repetitive‑pattern light to image defects). Determine whether it discloses elongated band/track filters and/or broadband/angularly diverse illumination. Flagged as unverified.

3. Non‑patent citations

3.1 George, N. & Jain, A., Applied Optics, "vol. 6," pp. 1202 ff. (June 1973)

• Posture: printed publication more than one year before filing → §102(b).
• Relevance: cited on the face and discussed in the SPEC as support for speckle reduction via spectral diversity — a multiple‑line argon laser, and a simulation using six laser lines spread over ~150 nm. It underpins the rationale (not the apparatus) of claim 1 and the SPEC's spectral‑diversity discussion. As a journal article on speckle, it cannot itself anticipate the apparatus claims; its proper use is §103 evidence that spectral diversity to suppress speckle was known and desirable.
• ⚑ Citation flag: the patent prints "vol. 6" (both on the face and in the SPEC). The commonly indexed version of this George & Jain paper appears as Applied Optics vol. 12, no. 6, pp. 1202–1212 (June 1973). I am reporting the identifier literally as it appears on the patent and flagging — not correcting — the apparent volume discrepancy; confirm before relying on the citation.

3.2 Thomas, C. E., Applied Optics, vol. 7, no. 3, p. 517 ff. (March 1968)

• Posture: §102(b) printed publication.
• Relevance: described in the SPEC as an early description of a device in which the spatial filter is rotated in the Fourier plane in synchronism with a rotating collimated light source, so that the dark spots track the diffracted spots — i.e., the prior‑art workaround for angular diversity using moving parts. Relevant to the angular‑diversity element of claim 14 and to the SPEC's criticism of "complex moving mechanisms." Not an apparatus anticipation (it is a moving‑filter system, not a static track filter).

4. Anticipation matrix — cited references vs. the independent claims

Legend: ● element present · ◐ partially present / arguable · ○ absent

Reading of the matrix: no row is all‑● for any independent claim.

5. Bottom line

1. No cited reference anticipates any independent claim (1, 14, 28, 38) under §102. Each lacks at least one of the two features that define the invention: (i) broadband and/or angularly diverse illumination, and (ii) a mainly transmissive Fourier‑plane filter bearing a plurality of spaced‑apart opaque tracks, each radiating from an imaginary point, matched to the elongated diffraction bands, feeding a 2D sensor. The references disclose the building blocks — spatial filtering of periodic patterns ('420, '280, '446), multiple sources ('252, '616), slit apertures ('420, '562), non‑laser/wavelength‑variable sources ('562), and the 2D‑array optical train ('774) — but not the combination.
2. Closest §102 candidates: US 3,658,420 (best against claim 28 and the filter‑centric claim 38) and US 3,790,280 (best against claim 38). Both are §102(b) art; both fail on "tracks radiating from an imaginary point," and '420 additionally fails the broadband limitation.
3. Two references carry elevated §102 risk via §102(e), not §102(a)/(b): US 5,172,000 and US 5,177,559. Because both were filed before '912 and issued after, they are easy to mishandle. US 5,172,000 in particular — "Spatial filter for optically based defect inspection system" — is the highest‑priority reference to obtain and read in full, given the subject‑matter match to claim 38.
4. Best §103 combinations (for completeness, since §102 is not the only exposure): {'774 or '446} as the architecture/2D‑sensor base + {'562 or '616 or '252} for non‑laser/multi‑source/slit/wavelength‑selective illumination + {'420 or '280} for blocking periodic‑feature energy, in view of George & Jain (1973) for the known use of spectral diversity to suppress speckle and Thomas (1968) for angular diversity. Two of these — '616 and '280 — cut against the invention on discrete features (wavelength‑selective filtering; planar‑filter defocus), so the §103 case would need careful articulation.

6. Verification gaps and flags (explicit)

• ⚑ Not retrieved this session (search‑step limit): full texts of US 4,806,774, US 4,895,446, US 5,172,000, and US 5,177,559. My element mappings for these four rest on '912's own BACKGROUND characterization and general knowledge and must be verified against the actual specifications before being relied on in any validity opinion. This is the single largest gap in this analysis.
• ⚑ Inventor names not verified for '420, '252, '774, '446, '000, and '559; only '280 (Heinz et al.), '562 (Anthon), and '616/others as noted above were confirmed.
• ⚑ Date inconsistency in the task environment: the task header states "Current Date: April 26, 2026," while the tool environment reports 2026‑09‑30. This does not affect the prior‑art dating above (all driven by 1992‑02‑07 and the references' own dates), but it is flagged for consistency with the previously generated sections, which used the April date.
• ⚑ Possible §112 observation (not §102): claim 24 recites source elevation angles "between 10° and 40°," while the SPEC states typical elevation angles "between 50° and 80° from normal incidence." These are mutually consistent only if claim 24 is read as measuring elevation from the substrate plane rather than from the normal (90 − 80 = 10; 90 − 50 = 40). Worth noting if claim 24 is ever construed.
• No auto‑correction applied: all patent numbers were treated literally as listed on the '912 face, and results for other "'912/491" numbers were discarded.

I'll ground the analysis in the prior art listed on this patent's page. Let me pull the key cited references to confirm what they actually disclose.

Let me confirm the two remaining key references cited on the page — the Bell Labs and Tokyo Shibaura (Minami) patents — since they anchor the broadband/coherent-illumination issue.

Obviousness Analysis — U.S. Patent No. 5,264,912

Statutory framework: Pre‑AIA 35 U.S.C. § 103 applies (application 07/832,379 filed 1992‑02‑07; granted 1993‑11‑23). Prior art therefore includes pre‑AIA § 102(a), (b), and (e) references. Critically, US 5,172,000 (filed 1990‑11‑02) and US 5,177,559 (filed 1991‑05‑17) are § 102(e) prior art because each was filed before — and granted on — the 1992‑02‑07 filing date. The 1976 patent US 3,971,562 below… note I mean US 3,981,562 and the 1968/1973 journal articles are classic § 102(b) art.

Scope of record used. Per instruction, I limit prior art to the references listed in this page's prior‑art section: the nine "Patent Citations" and two "Non‑Patent Citations." I supplemented titles with retrieved abstracts/passages to avoid mischaracterizing them; where I rely on a passage, I quote it. Where I could not retrieve full text, I say so.

1. Level of ordinary skill (PHOSITA)

As of February 1992: at least a B.S. in optics/applied physics plus ~3–5 years designing optical inspection systems, or a graduate degree with equivalent experience; may be a small team (an optical designer and a systems/metrology engineer). The PHOSITA is presumed to know (a) Fourier optics and the geometry of a transform plane, (b) intensity spatial filtering of periodic patterns, (c) that diffraction angle varies with wavelength via the grating equation reproduced in the '912 spec, (d) laser vs. arc vs. incandescent sources and their coherence properties, and (e) the well‑known phenomenon of laser speckle. Nothing in the claims requires anything beyond this level, and the '912 specification repeatedly describes its own contributions as applied, well‑understood optics ("Methods of designing such a lens or lens system are well known"; "Because means for reducing spatial coherence are well known, no further details … will be given").

2. The prior art of record, mapped to the claim elements

3. Independent‑claim analysis

Claim 1 — broadband source + "track" filter + 2D sensor

Element mapping:

• Inspection device for patterned wafers; periodic features diffracting into dispersion orders → Lin '774; US 4,895,446; US 5,172,000 all disclose this.
• Broadband source of illumination, beam along an optical axis → US 3,972,616 (broadband incoherent source) and US 3,981,562 (expressly non‑laser, tunable narrow‑band source chosen to "avoid the need for coherent light"). George & Jain supply the reason to broaden the spectrum.
• Aperture stop with the beam focused through it → Axelrod '420 (pinhole beam expander); Kodak '252 (aperture in an opaque plate); Anthon '562 (monochromator slit).
• Each order forming an elongated band → Anthon '562 literally produces "a multiple diffraction image of the monochromator slit" — i.e., an elongated (slit‑shaped) diffraction image; the '912 spec concedes the slit "causes each Fourier‑plane spot to spread into a strip, just as did the use of multiple colors."
• Mainly transmissive spatial filter in the Fourier plane with spaced‑apart opaque tracks blocking the bands, transmitting aperiodic scatter → US 5,172,000 ("array of … opaque stripes formed on a substantially transparent substrate"); Axelrod '420 (cross‑shaped opaque stop in the transform plane); US 3,790,280 (matrix array of opaque regions on a transparent field); US 4,895,446 (transform mask).
• Each track radiating from an imaginary point → geometry that follows from the transform plane (see §5).
• 2D imaging sensor + delivery means → Lin '774 (2D photodetector array); Axelrod '420 (TV camera + monitor); Anthon '562 (vidicon).

Proposed Ground 1: Lin '774 + Anthon '562 + George & Jain.

Motivation to combine, articulated as the case law requires:

1. Lin '774 and Anthon '562 are both in the same field (optical inspection of repetitive/periodic micro‑patterns) and address the same problem (suppressing periodic‑pattern diffraction so non‑periodic defects/particles stand out). KSR makes field‑of‑endeavor identity and the "interchangeability of known elements" a strong motivation.
2. Anthon '562 itself teaches away from lasers toward a non‑laser source and a slit geometry, and expressly notes it "avoids the need for coherent light" and "reduces the requirement for accurate alignment." A PHOSITA seeking to build a production patterned‑wafer inspector — where laser speckle and alignment tolerance are known pain points — would look to exactly this teaching.
3. George & Jain (1973) supply a reason to increase spectral bandwidth ("speckle reduction becomes more effective as the spectral bandwidth of the source is increased," per the '912 spec's own characterization) with a reasonable expectation of success (multi‑line argon; six lines over 150 nm).
4. Combining a broader/non‑laser source with the transform‑plane filter of Lin/'000 yields the predictable result that the formerly discrete order‑spots smear into bands — a result the '912 specification concedes is inherent ("If we remove the spectral filter and let all wavelengths be present simultaneously, then each former spot will appear as a band"). Shaping the opaque filter elements to match those bands is the predictable use of a known filter‑forming technique (photographic recording in/near the transform plane, per Scheff '000 and Axelrod/'4000949).

The "tracks" limitation is therefore the natural, predictable by‑product of pairing a known broadband/slit illumination scheme with a known Fourier‑plane blocking filter. No claim element requires a result beyond the ordinary skill level.

Claim 14 — multiple sources, each with its own aperture stop; filter blocks bands from all sources

Proposed Ground 2: Kodak '252 + Lin '774 (or '446) + George & Jain.

This is the strongest ground on the record because US 3,729,252 is nearly on all fours with claim 14:

• "plurality of sources … generating beams directed at a common region" → Kodak's multiple beams "directing … at said transparency," each "displaced from said optic axis";
• "each beam having an aperture stop to control its angular diversity" → Kodak's point sources "formed by a tungsten filament lamp positioned relative to each of said apertures," the apertures in an opaque plate fixing each beam's angular extent;
• "spatial filter … blocking said bands of light of said plurality of sources" → Kodak's light‑attenuating areas "in registry with a respective one of the light beams of zero‑order spatial frequency";
• beams "generally incoherent with respect to every other light beam" → Kodak claim 10, which is effectively the angular‑diversity/speckle‑reduction rationale of claim 14.

Kodak '252 even flags the exact trade‑off the '912 spec discusses — with multiple sources "the portion of the area of filter 26 which must be opaque may be nearly twice as large" (claim 14's specification analog), which the '912 spec resolves by keeping open area > ~50%. The remaining wafer‑inspection context comes from Lin '774/'446. Motivation: angular diversity for speckle reduction (Thomas '1968 sets up the problem; Kodak sets up the multi‑source solution; George & Jain corroborate diversity). Result predictable.

Claim 28 — single monochromatic source + slit‑shaped aperture stop + track filter

Proposed Ground 3: Lin '774 (or Axelrod '420) + Anthon '562 + Scheff '000.

• Monochromatic single source → Lin '774 (laser) / Axelrod '420 (He‑Ne laser).
• Elongated slit‑shaped aperture stop → Axelrod '420 discloses a "small rectangular slit" aperture in a laser system, expressly to increase S/N; Anthon '562 discloses a slit whose image is the elongated transform. A PHOSITA would substitute a slit for a pinhole to gain throughput and angular diversity — a design choice with predictable results (KSR; In re Aller's "obvious to try" / design‑choice line).
• Track filter → Scheff '000 / Axelrod '420.

Claim 28's use of a single monochromatic source with a slit removes any arguable overlap with the broadband cases and is squarely the subject of Axelrod '420 (slit + coherent laser + transform filter). The '912 spec's own justification — that a slit "permits the use of much more light from the source" and, "even when the source is a laser … the provision of angular diversity … can help to reduce speckle" — is a stated advantage of a known structure, i.e., a design choice, not patentable weight under KSR.

Claim 38 — the improvement: transmissive Fourier‑plane filter with opaque tracks radiating from an imaginary point

Proposed Ground 4: Scheff '000 (US 5,172,000) alone, or in view of Axelrod '420 and Heimz '280.

Claim 38 is drafted as a pure filter improvement. US 5,172,000 is § 102(e) art and discloses almost every element: a mainly transmissive filter (clear substrate) carrying multiple spaced‑apart opaque elements ("stripes") that block the periodic‑pattern diffraction orders while transmitting non‑periodic defect scatter, positioned in the Fourier transform plane, made photographically, and even indexed to the zero order with order‑dependent widths. The only facial gap is "radiating from an imaginary point" versus Scheff's "substantially parallel" stripes. That gap is closed by (i) Axelrod/'420 and Heimz/'280, which teach transform‑plane opaque patterns whose geometry is dictated by the specific feature/orientation distribution, and (ii) the ordinary‑skill knowledge that in an off‑axis or multi‑order geometry the order lattice originates at the specular (zero‑order) point — exactly the "imaginary point" the '912 spec later describes ("the apparent center … corresponds to the angle at which the specularly‑reflected beam leaves the substrate"). Where a claim recites an inherent geometric consequence of a known structure, In re Kao / In re Best inherency principles make the limitation unavailing as a patentability distinction.

4. Dependent claims

5. The pivotal limitation — "each track radiating from an imaginary point"

Because this phrase appears in all four independent claims, it is the limitation on which any validity defense would turn. It is the weakest point for the patentee:

• The '912 specification itself states the phenomenon is characteristic of such filters: "It is characteristic of spatial filters made in accordance with our invention that the opaque stripes appear approximately as segments of straight lines or curves radiating from an imaginary point … This phenomenon may be understood by recognizing that … each point in the Fourier plane represents a particular angle at which light is diffracted … The apparent center from which the stripes seem to radiate corresponds to the angle at which the specularly‑reflected beam leaves the substrate."
• That is a recitation of Fourier‑optic geometry, not an inventive act. Once illumination is spectrally diverse (spectrum runs radially outward from the zero/specular order) or angularly diverse (offset aperture shifts each order), the blocked bands necessarily emanate from the specular point. A limitation that is the inherent result of combining known elements does not confer patentability (In re Best; In re Kao). Scheff '000 (center stripe = zero order) plus Axelrod '420 (off‑axis-capable transform filter) supplies the structure; inherency supplies the geometry.

6. Teaching away and secondary considerations (counter‑arguments to weigh)

Strongest patentee counter‑argument — "teaching away." The '912 background characterizes US 3,972,616 as using the fact that "an ordinary spatial filter is ineffective for polychromatic light" as the basis of its operation. Read narrowly, '616 could be argued to teach away from spatial filtering of broadband light. But:

• KSR and In re Fulton require that a reference "criticize, discredit, or otherwise discourage" the claimed solution. '616 merely uses polychromatism for a different purpose (context imagery); it does not disparage band‑matched filters.
• US 3,981,562 (1976) — fourteen years before the '912 filing — directly discloses non‑laser, tunable, wavelength‑variable illumination used with a slit and a slit‑type transform filter to suppress a repetitive pattern's diffraction maxima. That reference undercuts any assertion that a PHOSITA believed filtering non‑laser light to be impossible.
• The '912 spec's admission that "It has not seemed possible until now to build an inspection machine which applies a spatial filter effectively to light from a broadband light source" is an admission of the state of the art that simultaneously (a) supports a "long‑felt need" argument for the patentee, but (b) is partially refuted by '562 and George & Jain, weakening the "unexpected result" story.

Secondary considerations. The record before me contains no evidence of unexpected results, commercial success tied to the claimed features, copying, or industry praise. Assignee success (Tencor/KLA) cannot be presumed to be nexus‑linked to these claims. Absent such evidence, the Graham secondary factors do not rescue the claims.

7. Bottom line

8. Caveats and confidence

1. Basis of the reference characterizations. I retrieved abstracts and key passages for US 3,658,420, US 3,729,252, US 3,790,280, US 3,981,562, US 5,172,000, and the '912 specification/EP family. I relied on the '912 specification's own description for US 3,972,616 (Minami) and US 4,895,446 (Maldari), and on the page's listing plus the George & Jain / Thomas citations for the two NPL items. Before filing any office action or invalidity contention, the full texts of '616, '446, and the two articles should be verified, clause by clause.
2. Anticipation vs. obviousness. Kodak '252 (claim 14) and Scheff '000 (claim 38) are close enough that a § 102 attack should be evaluated in parallel; I did not fully brief anticipation here because the task is § 103.
3. "Radiating from an imaginary point." I treat this as the sole potentially non‑disclosed feature for claim 38. If a tribunal rejected the inherency argument, claim 38 would survive on that narrow geometry alone — which is precisely why it is the claim to focus a rebuttal on.
4. Internal inconsistency flagged: claim 24 (10°–40°) vs. specification ("50° to 80° from normal incidence"). This is an independent vulnerability.
5. Not addressed here. The earlier sections noted the anomalous "Italimpianti of America" current‑assignee listing and the absence of litigation; neither affects the § 103 analysis. No litigation record was found, so there is no prosecution‑history estoppel or validity ruling to incorporate.

Confidence: High for the element mappings to Lin '774, Scheff '000, Kodak '252, Axelrod '420, Heimz '280, and Anthon '562 (quoted text). Moderate for Minami '616 and Maldari '446 (characterized via the '912 spec). The overall conclusion — that the independent claims are prima facie obvious over combinations of the cited art — is well supported; the residual uncertainty is the "radiating" limitation and the strength of the teaching‑away argument from '616.