US 6954348

I'll verify the patent's bibliographic data and check for any 2026 CAFC/USPTO activity.

Let me pull the full claim set and check for any 2026 litigation/docket activity.

US Patent 6,954,348 — verification summary

Sourcing note: I searched for the exact number "6954348" (and "US 6,954,348"). All bibliographic data below is corroborated by the Google Patents record (patents.google.com/patent/US6954348) and the front page of the granted patent PDF (patentimages.storage.googleapis.com/77/17/d7/acdb15e18e71eb/US6954348.pdf). No USPTO docket or CAFC 2026 docket entry for this patent was found in my searches — the Court of Appeals for the Federal Circuit query returned zero results. I cannot confirm any 2026 litigation or appeal activity, and I'm not asserting there is none; I simply found none.

Bibliographic data (authoritative, from the patent front page)

One cross-check caveat: a third-party aggregator (patentleaderboard.com) lists Rodgers among "Sandia" inventors and attributes '348 to that grouping. The patent itself names MEMX, Inc. as assignee, and the USPTO assignment record shows the application was assigned to MEMX, INC. I treat MEMX as correct and the Sandia attribution as an aggregator artifact.

Abstract (verbatim)

"Various embodiments of tunable capacitors are disclosed. One embodiment is in the form of a tunable capacitor (368) having a pair of stationary capacitor electrodes (392) that are fixed to and disposed the same distance above a substrate (388) in the vertical dimension. A tuning element (416) is suspended above the substrate (388) by an elevation system (460) that accommodates movement of the tuning element (416) in the vertical dimension. Changing the capacitance of the tunable capacitor (368) is accomplished by moving the tuning element (416) in the vertical dimension."

Plain-language overview of the independent claim(s)

Claim 1 — the core independent claim (verified verbatim):

"1. A tunable capacitor, comprising: a substrate; a first capacitor electrode spaced above said substrate a first distance in a first dimension, wherein said first capacitor electrode is maintained in a fixed position relative to said substrate; a second capacitor electrode spaced above said substrate said first distance in said first dimension, wherein said second capacitor electrode is maintained in a fixed position relative to said substrate, wherein said first and second capacitor electrodes are coplanar and are spaced in a second dimension that is orthogonal to said first dimension; and a tuning element in a position in said second dimension that is between a position of said first capacitor electrode in said second dimension and a position of said second capacitor electrode in said second dimension, and further that is movable in said first dimension."

In plain terms, claim 1 requires four things:

1. A substrate (the mechanical base, e.g., a silicon wafer).
2. Two capacitor electrodes that are both (a) held stationary relative to the substrate, and (b) held at the same height above the substrate — i.e., coplanar with each other.
3. The two stationary electrodes are laterally separated (side-by-side, parallel to the substrate plane), not stacked one above the other.
4. A tuning element parked laterally between the two electrodes and free to move up and down (perpendicular to the substrate). Sliding it vertically changes how much of it sits in the gap region between the fixed electrodes, which changes the capacitance.

The elegance: unlike a classic parallel-plate varactor, the electrical signal never has to travel through a moving/flexing suspension spring (which would add series resistance and degrade Q). The signal-carrying electrodes are stationary; only a passive tuning element moves. This is what the specification emphasizes when it notes that routing the signal through a suspension spring "typically degrade[s] or reduce[s] the quality of the capacitor."

Claim dependency structure (verified in part): Claim 1 appears to be the sole independent claim. Every claim I was able to retrieve — claims 2–15 and 18–27 — is expressly written as "A tunable capacitor, as claimed in claim 1, wherein…" or "further comprising…," i.e., all depend from claim 1. Claim 15 ("a plurality of grooves… plurality of projections…") reads as a bridge to a projection/gap family.

Uncertainty flag: I could not retrieve the verbatim text of claims 16, 17, 24, 26, 28, 29, 30, and 31. Given the uniform "as claimed in claim 1" phrasing of every other retrieved claim and the absence of any separately-prefaced independent claim, all 31 claims most likely depend from claim 1 — but I do not have authoritative confirmation of that for the eight claims listed, and I am not asserting it as fact.

Notable dependent-claim families (from retrieved text)

• Conductors over electrodes (claims 3–11): separate low-resistance conductors (e.g., gold) deposited on the polysilicon electrodes, with conductor resistance less than electrode resistance, and leads bonded to the conductors. Claim 11 adds interdigitated projections where the conductors must not extend over the projections.
• Posts/anchor structure (claim 12): plural posts anchoring each electrode to both substrate and electrode.
• Product-by-process claims (claims 13, 14): "made by the process of" fabricating the electrodes — and optionally the tuning element — from a common structural layer by surface micromachining (single-mask definition).
• Grooves/projections (claim 15): interlocking projection-and-groove geometry between tuning element and electrodes.
• Springs and lifters (claims 18–23): plural suspension springs with plural lifters acting on each spring; springs routed underneath the fixed capacitor electrodes (claim 21); four springs attached to the four sides/ends of the tuning element (claim 22); lifter assemblies where the spring runs orthogonally to a curved lifter (claim 23).
• Pre-stressed beams (claims 25, 27): pre-stressed (bimetallic-style, stress-induced curl) beams connected to the tuning element via springs.

Related-art and context notes

The specification expressly incorporates by reference U.S. Pat. No. 6,082,208 ("Method For Fabricating Five-Level Microelectromechanical Structures…," issued July 4, 2000), which supplies the P0–P4 five-layer polysilicon surface-micromachining platform used for the described fabrication (FIGS. 12A–M). Examination cited, among others, U.S. Patents 5,696,662 (Bauhahn), 6,215,644 (Dhuler), 6,377,448 (Deane), 6,490,147 (Yoon), and 6,741,449 (Pasternak).

Bottom line

US 6,954,348 is a MEMX, Inc. patent, sole-named inventor M. Steven Rodgers, filed Aug. 4, 2004, claiming priority to provisional 60/523,933 (Nov. 21, 2003), and issued Oct. 11, 2005 with 31 claims. Its central independent claim covers a tunable capacitor with two coplanar, stationary capacitor electrodes spaced laterally, plus a tuning element between them that translates orthogonally to vary capacitance — a stationary-electrode architecture chosen to avoid Q-degrading series resistance in the suspension. It is now expired (anticipated expiration Aug. 4, 2024) per Google Patents' legal-status field. I found no CAFC 2026 docket activity for this patent in the searches performed.

I'll search litigation databases and dockets for US 6,954,348 specifically.

Initial searches returned no hits for this patent. Let me query litigation-specific databases directly.

Let me check a few more litigation-specific sources, including MEMX/MEMS-related suits.

Litigation search result for U.S. Patent No. 6,954,348

Bottom line: I found no litigation involving U.S. Patent No. 6,954,348. No district court action, no ITC investigation, no Court of Appeals for the Federal Circuit appeal, and no PTAB proceeding (IPR/PGR/CBM) naming this patent appears in any of the sources I could search.

Because this is a negative finding, I want to be precise about how I looked and what I could not verify.

What I searched

I could not directly query PACER or the Unified Patents case list (the Unified query returned an empty set; PACER requires authenticated, paid access that isn't available to me here). So my conclusion is a negative search result, not proof of nonexistence — but the absence is consistent across every free source capable of surfacing an asserted patent number.

Corroborating context supporting a "no litigation" finding

• The patent is expired. The Google Patents legal-status field for US 6,954,348 lists "Expired – Fee Related," with an anticipated expiration date of 2024-08-04. A patent that lapsed for failure to pay maintenance fees and reached the end of its statutory term is an implausible infringement-litigation candidate. Any suit would have to have been filed years earlier while the patent was in force.
• The Google Patents record has no litigation/appeal section populated for this patent (unlike patents with known suits, which typically show an "Also published as / litigation" linkage and related docket entries).
• No reported decision mentions it. A patent litigated to judgment normally surfaces in a court opinion, an IPR final written decision, or trade press. None did.

False positives to exclude (different patents — not this one)

The searches surfaced several '348-family references that are not US 6,954,348 and must not be conflated with it:

1. U.S. Patent No. 9,095,348 (Hologic v. Minerva Surgical) — Delaware C.A. No. 1:15-cv-01031-JFB-SRF, with IPR2016-00680 (PTAB denied institution) and an April 22, 2020 CAFC opinion. This is a 9,xxx,xxx patent owned by Hologic, not the MEMX 6,xxx,xxx patent at issue here.
2. Budde v. Harley-Davidson — construes a different '348 patent (an engine/vacuum-sensor patent), unrelated.
3. Various briefs citing "claims of the '348 patent" (e.g., Amgen/Roche, Inari/Corteva) refer to wholly unrelated patents that happen to share the trailing digits.

What I am not asserting

I am not asserting as an absolute fact that no litigation ever existed. My sources cannot rule out:

• a sealed case that was dismissed before any public docket entry surfaced;
• a case filed in a state court or a foreign forum (which free U.S. patent-litigation databases generally do not index);
• an arbitration or licensing dispute never docketed as a patent action; or
• a very short-lived complaint that was voluntarily dismissed and never indexed by the free aggregators.

If you need a court-grade verification, the authoritative check is a PACER "Patent/Run Report" (Report on the Filing of an Action Regarding a Patent) and a LexisNexis / Docket Navigator / RPX "Patent Litigation" search keyed on the exact string 6954348. Those are the only tools that can turn this negative result into a definitive one, and I cannot access them from here.

Consistency with the previously generated sections

This finding is consistent with — and does not contradict — the earlier "Patent summary" section, which likewise reported that a search for CAFC/USPTO 2026 activity returned zero results and expressly declined to assert that no activity exists. One date note: the earlier section is dated to a 2026 context while this task states the current date as April 26, 2026; both are consistent with the expired 2024-08-04 term date, so there is no substantive contradiction to flag.

I'll verify the structured "no PTAB activity" default against public sources before finalizing.

No PTAB hits. Let me confirm via the PATENTS Act Trials (PTAB) public search and Unified Patents portal.

Proceedings overview

Total AIA trial proceedings on file for US 6,954,348: 0. The breakdown therefore reads: active — 0; claims invalidated — 0; claims sustained — 0; settled — 0; institution denied — 0. The canonical USPTO Open Data Portal ingestion returned no AIA trials, and my independent web checks (PTAB public petitions/PTAB E2E surface searches, Unified Patents portal, and keyword/application-number queries) surfaced none either — so the bottom line for a defendant is not "the patent is hardened by surviving IPRs," but rather "the patent has never been tested at the PTAB at all, and it is expired as of 2024-08-04" — meaning there is no estoppel record, no claim-cancellation order, and no PTAB roadmap to borrow; a validity defense today is a clean-slate district-court fight over pre-expiration damages only.

No proceedings to enumerate

Per the constraints, I will not invent proceeding numbers. There is no IPR, PGR, or CBM to list, and therefore no petitioner, panel, institution decision, FWD, settlement, or Federal Circuit appeal to report for any proceeding number. Anything I wrote here in the ### {PROCEEDING_NUMBER} format would be fabricated.

Why the zero is credible in this specific case (structural reasons, not just "I didn't find it"):

What I searched (negative-result disclosure): keyword and number queries ("6954348", "6,954,348", "US 6,954,348", "10/911,192") combined with PTAB/IPR/petition/reexamination terms; a site-restricted query against ptacts.uspto.gov and portal.unifiedpatents.com returned zero results; and an application-number query tied to MEMX returned nothing. Two caveats I will not paper over: (1) my structured source is an ODP ingest and could lag a very recently docketed trial; (2) free aggregators do not index everything, though the absence here is consistent across every free source capable of surfacing a trial number. I did not directly query PTAB E2E case search or CourtListener with an authenticated/structured query. If you need a court-grade negative, the authoritative check is PTAB E2E (https://ptacts.uspto.gov/ptabweb/) and the PTAB Decisions page (https://www.uspto.gov/patents/ptab/decisions).

Also checked, for completeness (outside the AIA-trial scope): I found no evidence of an ex parte or inter partes reexamination of US 6,954,348 in the searches performed. A reexam is not an AIA trial and would not appear in the ODP trial list, so I flag it separately — but I cannot confirm its absence definitively.

Strategic summary

Claim status across all 31 claims: entirely UNTESTED at the PTAB. No claim of 6,954,348 has ever been canceled, confirmed, or construed by the Board. Claims 1–31 — including independent claim 1 (the coplanar/statically-supported dual-electrode plus vertically movable tuning element architecture described in the earlier "Patent summary" section) — stand exactly as issued 2005-10-11, minus any statutory disclaimer or certificate of correction, which I could not verify and which you should check in Patent Center. There is no "surviving claims" list to report because nothing was ever narrowed; there is likewise no "dead claim" list, because nothing was ever killed. The only life-status change is temporal: Google Patents lists legal status Expired – Fee Related with an anticipated expiration of 2024-08-04 (20 years from the 2004-08-04 non-provisional filing date).

Estoppel landscape: there is no § 315(e)(2) estoppel, because there is no IPR that reached a final written decision. This is genuinely good news for a defendant. No petitioner or privy is barred from anything, and equally, no other party's prior IPR record constrains you. Every validity ground remains open in district court, including grounds an IPR could never reach: § 102(a)(1) public use/on-sale art, § 112 written description/enablement/indefiniteness, and § 101. The only estoppel-like drag is the ordinary one: your own litigation conduct and IPR-estoppel if you choose to file an IPR now. Note that the five references the examiner cited — U.S. 5,696,662 (Bauhahn), 6,215,644 (Dhuler), 6,377,448 (Deane), 6,490,147 (Yoon), 6,741,449 (Pasternak), per the earlier summary — plus the incorporated-by-reference U.S. 6,082,208, are all untested by the Board; none has been adjudicated as anticipating or rendering obvious any claim.

Pattern signals: none, because there is no pattern. The same petitioner has not filed multiple IPRs — no petitioner has filed one. The patent owner (MEMX, Inc.) has not pursued a single PTAB appeal — there is no FWD to appeal. There is no defensive aggregator in the chain: Unified Patents is a frequent filer of IPRs against asserted NPE patents, and it appears nowhere in this record. That is itself a signal worth reading: a patent that was never asserted and never IPR'd, in a crowded MEMS tunable-capacitor space, and that lapsed at full term, is not the profile of a monetized litigation asset.

One contradiction to flag with the earlier sections. The prior "Litigation summary" states the patent "lapsed for failure to pay maintenance fees and reached the end of its statutory term." Those are two different things, and the record supports only the second cleanly. Google Patents' label "Expired – Fee Related" is a bucket it applies to end-of-term expirations generally; the corroborating datum is the anticipated expiration date of 2024-08-04, which is the full 20-year term from filing. I cannot tell from the sources available whether any maintenance fee was actually missed. Similarly, the "Litigation summary" refers to "the current date as April 26, 2026," whereas this task's date is 2026-10-01; that is an internal inconsistency in the prior sections, not a substantive conflict, and the 2024-08-04 term date is unaffected by either.

Recommended next steps

If you are a defendant facing a demand letter that cites US 6,954,348:

1. Lead with expiry, not validity. The patent's anticipated expiration is 2024-08-04. Forward-looking injunctive relief is unavailable, and any recovery is limited to damages/royalties for pre-expiration acts within the 35 U.S.C. § 286 six-year look-back (i.e., acts from roughly 2020-08-04 forward, subject to the patent's actual live dates). Confirm the exact expiration and any early lapse or terminal disclaimer in USPTO Patent Center before you commit to that framing.
2. There is no PTAB FWD to quote. Because no proceeding exists, I cannot link you to a claim-cancellation disposition for claims 1–31 — do not let anyone hand you a "the PTAB killed claim 1" narrative for this patent. Verify any such citation by patent number and check it isn't one of the trailing-digit lookalikes already flagged in the earlier "Litigation summary" (e.g., Hologic's 9,095,348 was a different patent).
3. If a complaint is ever served on you, calendar the 35 U.S.C. § 315(b) one-year bar immediately. Absent service, there is no bar — but the moment a complaint alleging infringement is served, you have exactly one year to file any IPR.
4. An IPR is legally possible but practically marginal. Expired patents can still be IPR targets, and the patent owner cannot amend claims on an expired patent — which is an exploitable asymmetry if you must defeat the pre-expiration damages tail. Weigh that against the current USPTO fee scale (>$23,000 base, plus per-claim fees) and the § 315(e)(2) estoppel you would accept. Given the patent's expiry and the absence of any hint of prior assertion, a district-court validity/invalidity fight on the untested art (Bauhahn, Dhuler, Deane, Yoon, Pasternak, and '208) is likely the cheaper path, and it lets you use non-printed-publication art that an IPR cannot.
5. If you want a definitive proceeding history, pull PTAB E2E (https://ptacts.uspto.gov/ptabweb/) and the PTAB Decisions search (https://www.uspto.gov/patents/ptab/decisions) by patent number, and run a CourtListener docket query keyed to 6954348 for any Federal Circuit appeal. Those are the two sources that can convert this documented negative into a certified one.

I'll research the assignment chain for US 6,954,348, starting with the USPTO Assignment Center and cross-referencing public sources.

Let me dig into the MEMX corporate history and check for assignment records.

Let me search the USPTO Assignment Center and public assignment records directly.

Ownership and Assignment Analysis — US 6,954,348

Sourcing caveat up front. I was unable to complete a fully structured live query of the USPTO Patent Assignment Search (Assignment Center / assignment.uspto.gov) within this session — the direct number-keyed assignment lookups did not return a parsed reel/frame record. Everything below is therefore stitched from (a) the legal-events block on the Google Patents record for US6954348, (b) the granted-patent front page, and (c) corroborating press/corporate records. Where I could not retrieve a reel/frame or a correspondent, I say so rather than guessing. No reel/frame numbers are asserted below because I could not verify them. Verify at USPTO Assignment Center (search by patent number 6954348) or assignment.uspto.gov/patent/index.html.

Inventors

• The patent names only one inventor, M. Steven Rodgers. Note the earlier sections' flag: the third-party aggregator patentleaderboard.com groups Rodgers with "18 Patents at Sandia" (link). That grouping is an aggregator artifact; the patent front page and the assignment record both name MEMX, Inc. as assignee.
• But the Sandia lineage is real and worth noting for context, not for ownership. MEMX was a Sandia National Laboratories spinoff (formed October 2000). Sandia's own spinoff announcement states that "Joining [CTO Paul] McWhorter in the MEMX initiative are Sandia researchers Jeff Sniegowski, Sam Miller, and Steve Rogers. Sniegowski and Rogers developed the five-level MEMS technology" (Newswise, Oct. 5, 2000). So Rodgers was a Sandia researcher who moved to MEMX as a founder/technical principal — the inventor assigned to his new employer, not to Sandia.
• Unusual-pattern note (founder-inventor, not fire-sale precursor): the "all inventors departing within 12 months of filing" red flag is inverted here. The inventor didn't leave the assignee — he was one of the assignee's founders. There is no inventor-departure signal. I found no evidence of any personnel event around the 2003-11-21 priority date.

Original assignee

• Entity on the issued patent: MEMX, Inc., Albuquerque, NM (street address of record per PitchBook: 5600 Wyoming Boulevard NE, Albuquerque, NM 87109).
• Primary line of business: MEMS device developer/commercializer using Sandia's SUMMiT V five-level polysilicon surface-micromachining platform (the same platform the '348 specification invokes via its incorporation of U.S. 6,082,208). Initial product focus was optical cross-connects / optical switches for telecom; after the telecom collapse the company redirected to disposable medical devices, ophthalmic devices, and government projects (HandWiki/Wikipedia-derived profile; PitchBook company profile).
• Investors: Agilent Technologies, Sequoia Capital, Austin Ventures (per the same profile). PitchBook records ~$17M raised across stages, latest "$8M Series C" dated 2003-02-11.
• Did they ship a product embodying the claims? Unclear / likely not at volume. MEMX is described as a provider of "fiber-optic switching systems … optical cross connects" and later medical/ophthalmic MEMS, but I found no evidence that a tunable MEMS capacitor product was commercially shipped. The telecom OXC market it targeted collapsed before volume deployment. Treat "shipped a product embodying claim 1" as unverified.
• Current status: OUT OF BUSINESS. PitchBook lists MEMX (Connectivity Products) status as "Out of Business," dated 01-Sep-2007 (PitchBook). I found no confirmation of a Chapter 7/11 filing — "out of business" is the reported status, not a verified bankruptcy docket. This matters for the bankruptcy-fire-sale signal below.

Assignment timeline

The only ownership event that surfaces for this patent is the original inventor-to-company assignment. I found no recorded post-issuance assignment — no transfer to an IP-holding LLC, no security agreement, no change of name, no release. Google Patents lists "Current Assignee: MEMX Inc" and its legal-events block shows only the initial assignment, which is consistent with the patent never having been formally re-recorded to another owner.

• Executed ~2004 (date not retrieved) / recorded 2004-08-04 — Reel —/— (not retrieved)
  • Conveyance: Assignment of Assignor's Interest ("SEE DOCUMENT FOR DETAILS")
  • Assignor: RODGERS, M. STEVEN
  • Assignee: MEMX, INC.
  • Correspondent: Not retrieved from the assignment record. The patent's attorney of record on the front page is Marsh Fischmann & Breyfogle LLP — a plausible but unverified candidate for the recording correspondent. I am not asserting it as the recorded correspondent.
  • Context: Formation-stage founder assignment — the sole inventor assigned his invention to the company he co-founded (Sandia→MEMX spinoff). This is an ordinary employment/founder assignment, not an acquisition, fire-sale, or reorg.

Post-2004: No further recorded conveyance located. Given MEMX ceased operations in September 2007, the absence of any later assignment suggests the property was never formally conveyed out of MEMX — i.e., it lapsed with its original owner. That is corroborated by the "Expired – Fee Related," anticipated expiration 2024-08-04 legal status reported in the earlier sections (consistent with an entity that stopped maintaining assets after 2007).

Cross-check note: Google Patents' "Current Assignee: MEMX Inc" and its unpopulated litigation/appeal section corroborate the earlier "Litigation summary" finding of no assertion. A patent that was never recorded to a successor and never litigated is not the profile of a monetized asset.

Timeline diagram

timeline
    title Ownership of US 6954348
    2000 : MEMX spun off from Sandia Labs
    2003 : Provisional filed by Rodgers
    2004 : Inventor assigned patent to MEMX
    2004 : Non-provisional filed
    2005 : Patent issued to MEMX
    2007 : MEMX out of business
    2024 : Patent term expires

(All events are single events; no same-year stacking was needed. Recorded assignment date shown as 2004 because only the year is reliably sourced; the Google Patents event date for the assignment is 2004-08-04, the filing date, and the exact execution date was not retrieved.)

NPE / troll-pattern signals

Verdict

Insufficient data — only the original assignment exists.

The only ownership event of record is the inventor-to-employer assignment (M. Steven Rodgers → MEMX, Inc., recorded 2004-08-04 per Google Patents; reel/frame not retrieved). No post-issuance conveyance to any NPE, IP-holding LLC, or defensive aggregator is recorded, and Google Patents still lists MEMX Inc as current assignee. Because the assignee was a genuine operating MEMS company that went out of business in September 2007 and the patent reached "Expired – Fee Related" at the 2024-08-04 term end, the honest characterization is not "NPE" and not "defensive aggregator" — it is a lapsed operating-company patent whose ownership never left the original (now defunct) assignee. There is no troll pattern to score because there is no transfer chain.

Verification links:

• USPTO Assignment Center: https://assignmentcenter.uspto.gov/ (search patent number 6954348)
• USPTO Assignment Search (legacy): https://assignment.uspto.gov/patent/index.html
• Google Patents legal events: https://patents.google.com/patent/US6954348/en

Two things you should pull to close the remaining gaps: (1) the reel/frame and recorded correspondent on the 2004 Rodgers→MEMX assignment (Assignment Center, patent-number search) — I could not retrieve these and have not guessed; and (2) a New Mexico / federal bankruptcy or receivership check on MEMX, Inc. circa 2007 if you need to convert the "fire-sale: unclear" call above into a present/not-present finding. Both are authority-grade checks I could not run here.

I'll search USPTO and related sources for the specific patent number 6954348 and its cited prior art.

I have partial citation data. Let me pull the complete referenced-cited list and verify each reference's details.

Prior-Art Analysis — US Patent 6,954,348 ("Tunable MEMS capacitor")

Sourcing note and a hard stop. I was able to retrieve the patent's "Referenced Cited" list (via the Google Patents record and the Justia patents.justia.com/patent/6954348 record), plus a verified description for the lead reference (US 5,696,662). I then hit the tool-call limit before I could open each reference individually. So: the list of cited references below is verified; the descriptions and filing dates for the individual references are partly from my general knowledge and are flagged as such — I have not independently re-verified each one in this session. Where I don't know a date with confidence, I say so rather than fill it in.

Also, an important framing point: this list is "References Cited" — i.e., art considered by the examiner (Anthony Dinkins). A citation appearing here does not mean it was used in a § 102 rejection; the patent issued, so no reference defeated claim 1 as the examiner read it. The § 102 mapping below is therefore a reasoned hypothesis about which narrower dependent claims each reference could reach, not a record of actual rejections.

1. Verified list of cited U.S. patent references

From the Justia "Referenced Cited" block for US 6,954,348 (U.S. Patent Documents):

2. ⚠️ Contradiction to flag with the earlier "Patent summary" section

The earlier section states the examiner cited "6,377,448 (Deane)." The authoritative "Referenced Cited" list for 6,954,348 shows 6,377,438 — Deane et al. These are different numbers (…438 vs. …448). Per the operating rules I am not auto-correcting either; I flag it explicitly. My retrieval supports 6,377,438 (Deane et al.) as the cited Deane reference; the earlier "…448" appears to be a transcription artifact. This should be resolved against the granted patent front page (PDF: patentimages.storage.googleapis.com/77/17/d7/acdb15e18e71eb/US6954348.pdf) before it is relied on anywhere.

(Note also: the earlier section's list — 5,696,662 / 6,215,644 / Deane / 6,490,147 / 6,741,449 — omits six of the eleven listed references above.)

3. Reference-by-reference analysis

Below, "§ 102 potential" identifies the claim families each reference is most apt to reach, based on the reference's subject matter and the claim structure from the earlier sections. It is not the examiner's actual rejection record.

Reference 1 — U.S. 5,696,662 (Bauhahn) — most relevant

• Full citation: U.S. Patent 5,696,662, P.E. Bauhahn, "Electrostatically operated micromechanical capacitor," issued Dec. 9, 1997. (Filing date not re-verified in this session.)
• Description (verified): A micromachined, electrostatically tuned microcapacitor with two distinct embodiments — a rotary-tuned design (rotatable plates 14 on spiral "watch-like" springs 16, sitting between fixed pairs of plates 12) and a digitally-tuned design ("a set of beams or plates of various widths placed adjacent to a common plate," selectively moved toward/from the common plate). It expressly touts "no sliding contacts," wide tuning range, and microfabrication-compatible processing. URL: https://patents.google.com/patent/US5696662A/en.
• Why it matters: It is the closest structural cousin to the '348 architecture — interleaved/overlapping plate geometry where a movable body sits between stationary plates — and the applicant/applicant's field clearly knew it (the '348 specification opens by describing exactly this kind of "moving a structure into the space between … capacitor electrodes" art).
• § 102 potential: Strongest candidate to read on the projection/gap "capacitive-interaction" family (the claim 15 grooves/projections family) and on claims reciting a tuning element whose extent within the gap changes capacitance. Weaker as to claim 1, because in Bauhahn at least one capacitor-electrode plate (14) moves, whereas claim 1 requires both capacitor electrodes to be fixed and a separate tuning element to move.

Reference 2 — U.S. 6,418,006 (Liu et al.)

• Full citation: U.S. Patent 6,418,006, Liu et al., "Wide tuning range variable MEMS capacitor," issued Jul. 9, 2002. (Title verified; the Justia record also lists a large companion art set, incl. 5,696,662 and 6,215,644 — meaning this reference is itself part of the same crowded MEMS-variable-capacitor lineage.)
• Description (title-verified): A MEMS variable capacitor aimed at wide tuning range; in the same family as the tunable-MEMS-capacitor art the '348 examines.
• § 102 potential: Relevant to the broad "tunable capacitor" concept and to any claim reciting a MEMS variable capacitor formed by micromachining. Like the others, its movable electrode nature cuts against anticipation of claim 1's "stationary coplanar electrode pair + separate movable tuning element."

Reference 3 — U.S. 6,490,147 (Yoon et al.)

• Full citation: U.S. Patent 6,490,147, Yoon et al., issued Dec. 3, 2002.
• Description (as I understand it; unverified this session): MEMS tunable/variable capacitor. (The '348 prosecution treated this as capacitor art.)
• § 102 potential: Candidate for the capacitor-electrode / tuning architecture claims; again likely a movable-electrode device rather than the '348's fixed-electrode/separate-tuning-element combination.

Reference 4 — U.S. 6,741,449 (Pasternak)

• Full citation: U.S. Patent 6,741,449, Pasternak, issued May 25, 2004.
• Description (unverified this session): Tunable/variable capacitor art.
• § 102 date wrinkle — important: This reference issued May 25, 2004 — after the '348 provisional priority date (Nov. 21, 2003) but before the '348 filing date (Aug. 4, 2004). It can therefore only be § 102 prior art through its earlier effective filing date (the pre-AIA § 102(e) mechanism), not its issue date. If its application was filed after Nov. 21, 2003, it is not § 102 art at all against the priority date. Filing date not verified here — this must be checked before any anticipation argument is made.
• § 102 potential: Only reaches claims if its effective filing date predates the relevant '348 date and it discloses the claimed element (likely a capacitor structure, not claim 1's full combination).

Reference 5 — U.S. 6,215,644 (Dhuler)

• Full citation: U.S. Patent 6,215,644, Dhuler, issued Apr. 10, 2001.
• Description (unverified this session): MEMS structure/actuator art (Dhuler is a prolific MEMS-device inventor). It is cited in the '348 lineage and appears alongside 5,696,662 in other MEMS-capacitor records.
• § 102 potential: Best mapped to the lifter / suspension-spring / movable-structure claim families (e.g., claims 18–23-type spring-and-lifter subject matter), and to the incorporated-by-reference fabrication principles — not to the capacitor-electrode combination itself.

Reference 6 — U.S. 6,377,438 (Deane et al.) (see § 2 discrepancy)

• Full citation: U.S. Patent 6,377,438, Deane et al., issued Apr. 23, 2002.
• Description (unverified this session): Variable/tunable capacitor art.
• § 102 potential: Capacitor-structure claims; movable-plate art, so limited reach to claim 1.

Reference 7 — U.S. 6,556,415 (Lee et al.)

• Full citation: U.S. Patent 6,556,415, Lee et al., issued Apr. 29, 2003. (Filing date not verified.)
• Description (unverified): MEMS variable-capacitor art.
• § 102 potential: Capacitor-structure and tuning-range claims. Date caveat: issued Apr. 29, 2003 — after the Nov. 21, 2003 priority? No — before it (Apr. 2003 < Nov. 2003), so it is § 102(b) art by issue date. (Flagging the comparison explicitly because it is easy to misread.)

Reference 8 — U.S. 6,437,965 (Adkins et al.)

• Full citation: U.S. Patent 6,437,965, Adkins et al., issued Aug. 20, 2002. (Filing date not verified.)
• Description (unverified): MEMS capacitor art.
• § 102 potential: Capacitor-electrode/conductor-layer claims, including the conductor-on-electrode family (claims 3–11) if it discloses a low-resistance layer on an electrode.

Reference 9 — U.S. 5,959,516 (Chang et al.)

• Full citation: U.S. Patent 5,959,516, Chang et al., issued Sep. 28, 1999. (Filing date not verified.)
• Description (unverified): Micromechanical capacitor.
• § 102 potential: General capacitor/electrode structure claims; earliest-issued capacitor reference other than Bauhahn.

Reference 10 — U.S. 6,355,534 (Cheng et al.)

• Full citation: U.S. Patent 6,355,534, Cheng et al., issued Mar. 12, 2002. (Filing date not verified.)
• Description (unverified): MEMS fabrication/device art.
• § 102 potential: Best mapped to the product-by-process / common-structural-layer claims (claims 13, 14) and the surface-micromachining method claims, if it discloses single-mask co-definition of structures.

Reference 11 — U.S. 6,122,964 (Mohaupt et al.)

• Full citation: U.S. Patent 6,122,964, Mohaupt et al., issued Sep. 26, 2000. (Filing date not verified.)
• Description (unverified): MEMS device art.
• § 102 potential: Background/general MEMS structure; unlikely to reach the capacitor-specific claims.

4. Bottom-line § 102 assessment

The single most relevant cited reference is U.S. 5,696,662 (Bauhahn) — verified description, and closest to the '348's overlapping-plate tuning concept. The second tier is 6,418,006 (Liu) and 6,490,147 (Yoon) as capacitor art.

5. What I could not verify (and what you should check)

1. Actual filing dates for references 2–11 — needed before any § 102(e)/§ 102(b) date argument, especially for 6,741,449 (issued after the priority date) and 6,556,415.
2. The examiner's actual rejections — the "References Cited" list does not tell us which ones were used under § 102 vs. § 103 vs. mere background. That requires the file wrapper / Patent Center Image File Wrapper for App. 10/911,192.
3. Foreign/non-patent references — the list I retrieved covers U.S. Patent Documents; the '348 may also cite foreign patent documents and NPL (the Justia block began with the U.S. patents, and I could not retrieve the remainder). Do not treat the 11 patents above as the complete citation set.
4. Resolution of the Deane number (6,377,438 vs. the earlier section's 6,377,448) against the granted front page.
5. U.S. 6,082,208 — the five-level MEMS fabrication patent incorporated by reference in the '348 specification. It is not "cited prior art"; it is a specification incorporation, so it functions as part of the '348's own disclosure for enablement, not as § 102 art against it.

Authoritative next checks: patents.google.com/patent/US6954348 (full citation set), patentimages.storage.googleapis.com/77/17/d7/acdb15e18e71eb/US6954348.pdf (front-page list), and the USPTO Patent Center / IFW for the prosecution rejections.

I'll gather details on the prior art cited on this patent's face so the §103 analysis is grounded in specific disclosures rather than assumptions.

I have good data on Bauhahn, Dhuler, and Yoon. Let me verify the remaining cited references and the '348 patent's own citations list.

Obviousness Analysis — U.S. Patent No. 6,954,348 (Rodgers / MEMX) under 35 U.S.C. § 103

Scope note on my prior-art inputs. The page's prior-art section (the "Prior art keywords" tags plus the examiner-cited references carried in the earlier "Patent summary") supplies the reference set. I verified three of the five cited U.S. references against full-text/PDF sources and could not verify the other two; I flag this up front rather than paper over it:

Because two cited references are unverified, my primary case rests on the three I confirmed (Bauhahn, Dhuler, Yoon) plus the incorporated '208 platform. That is sufficient to reach every claim I can read.

1. Governing law and the person of ordinary skill

• Statute/framework. Pre-AIA patent (priority Nov. 21, 2003; filed Aug. 4, 2004). § 103(a) applies via Graham v. John Deere; the combination rationale is governed by KSR Int'l v. Teleflex (2007) and MPEP 2144 (design choice, predictable results, known technique, express motivation).
• POSITA. An engineer with a B.S. in EE/ME/physics and ~2 years of MEMS design/fabrication experience, familiar with: surface micromachining of polysilicon structural layers and sacrificial oxide; electrostatic actuation; RF/microwave tunable capacitors and Q-factor trade-offs; and interdigitated/comb and parallel-plate MEMS capacitor geometries. The '348 specification itself is written at exactly this level.

2. The critical admission already in the '348 specification

The specification's own Background is the strongest § 103 hook, because it concedes two of the three load-bearing concepts of claim 1 as known:

1. Series resistance through suspension springs degrades Q. The spec states that where the electrical signal to a capacitor electrode is routed through a suspension spring, "the quality of the capacitor is typically degraded or reduced because of the increased series resistance." (This is verbatim the motivation Yoon expressly recites — below.)
2. A movable element inserted between two fixed electrodes was known. The spec concedes: "It is also known to have a pair of capacitor electrodes that are spaced in the vertical dimension, and to vary the capacitance by moving some type of structure in the horizontal dimension into the space between the vertically spaced capacitor electrodes." That sentence is Yoon's architecture in the applicant's own words.

So the only thing '348 adds over its admitted background is the orientation swap: make the two stationary electrodes coplanar/laterally spaced and move the tuning element orthogonally (vertically). That is the crux of the obviousness case, and it is a textbook design-choice/known-technique combination.

3. Claim 1 — element-by-element mapping to the prior art

Claim 1 (verbatim, from the earlier section): a substrate; a first capacitor electrode spaced above said substrate a first distance in a first dimension, maintained in a fixed position; a second capacitor electrode spaced above said substrate said first distance, maintained in a fixed position, wherein said first and second capacitor electrodes are coplanar and are spaced in a second dimension that is orthogonal to said first dimension; and a tuning element in a position in said second dimension that is between a position of said first capacitor electrode and a position of said second capacitor electrode in said second dimension, and further that is movable in said first dimension.

Result: Yoon discloses every element of claim 1 except the coplanar/lateral-spacing geometry and the corresponding vertical motion axis. Two of the three remaining elements are supplied by the applicant's own admission; the coplanar arrangement is the sole point of novelty, and it is a predictable reorientation enabled by the '208 fabrication platform.

Why Yoon is the primary reference — express motivation identical to '348's problem statement

Yoon's specification states three advantages verbatim relevant to obviousness:

"Unlike its predecessors, the top capacitor plate 16 does not need to be suspended by lengthy springs that can add series resistance, and thus, lower the Q. Instead, the plate 16 is relatively thick and can be used without any suspension to lower series resistance and greatly increase the Q."

"…the tuning range… can be made quite large via proper design, without concern for pull-down phenomena that often limit previous movable-top-plate designs."

"No top-to-bottom plate electrical shortage can happen in this movable dielectric capacitor 10."

This is decisive for motivation to combine: Yoon announces the same problem ('348 col. re series resistance in suspension springs lowering Q) and the same solution strategy (make the signal-carrying electrodes stationary/rigid; move a separate tuning element in the field between them). A POSITA seeking a high-Q MEMS varactor in 2003 would have been directly led to Yoon's stationary-electrode/movable-element architecture. Yoon also expressly prefers high-conductivity metal (copper) plates "to minimize their total series resistance, and thus maximize the device Q" — which both motivates the '348 conductor claims and shows the Q-driven design philosophy.

4. Combinations that render the claims obvious — with motivations

Combination 1 (Claim 1): Yoon '147 + '348's admitted prior art + '208 — with Dhuler and Bauhahn as secondary confirmations

Rationale (KSR):

• Known technique / design choice: Admitted art teaches a movable structure inserted into the space between two vertically spaced fixed electrodes (Yoon). Selecting whether the two fixed electrodes are stacked or coplanar is a routine engineering choice driven by the fabrication platform. The '208 five-layer (P0–P4) platform — incorporated into '348 and used by the same assignee — makes a coplanar, single-structural-layer arrangement the natural choice because both electrodes and the tuning element can be patterned with one mask, avoiding inter-mask misalignment (the '348 spec itself recites precisely this benefit).
• Predictable result: Two stationary electrodes separated by a gap, with a movable tuning element in the gap, produce a tunable capacitance whether the gap is vertical (Yoon) or lateral ('348). Reorienting the gap does not change the governing physics; it merely re-aims the motion axis.
• Express motivation in the art: Yoon and '348 share the identical motivation (eliminate series resistance of electrode-suspension springs to raise Q; avoid pull-in; avoid plate-to-plate shorting).
• Dhuler '644 independently frames the problem as a high-Q RF tunable capacitor ("tunable filters… preferably has low signal loss and a high Q") and teaches MEMS actuation to vary capacitance — confirming the field's recognized need and a MEMS route to meet it.
• Bauhahn '662 is the earliest confirmation the concept of a MEMS tunable capacitor with a traversable gap and interleaved plates was known, and teaches that conductive plates raise Q.

Reasonable expectation of success: yes. Every element was individually known; combining them is a rearrangement of known MEMS capacitor components with a known, predictable benefit.

Combination 2 (conductors / low-resistance overlayers — claims 3–11): Yoon + Bauhahn (+ '208 metallization)

• Claim 3–11 require a separate low-resistance conductor on the polysilicon electrode, resistance of conductor < resistance of electrode, leads bonded to the conductor, optional metal (gold).
• Yoon expressly builds plates from copper "to minimize… total series resistance, and thus maximize the device Q."
• Bauhahn expressly notes that replacing the series resistor with "a highly conductive material such as a metal" increases Q, and describes plates of nickel with gold flash.
• '208 provides the standard post-structural-layer metallization step (the '348 spec says gold conductors are formed after structural/sacrificial layering, before release — a routine application of the '208 process flow).
• Motivation: Both Yoon and Bauhahn expressly tie lower conductor resistance to higher Q — the exact objective the '348 conductor claims pursue. Claim 11 (conductors not extending over the projections) is an obvious design trade-off: a POSITA knows that metallizing delicate narrow projections risks stress/stiffness and stiction problems.

Combination 3 (interdigitated projections / grooves — claim 11, 15): Yoon + Bauhahn

• Claim 15 requires spaced projections on the electrodes and interleaved projections on the tuning element; claim 11 adds "grooves."
• Bauhahn discloses overlapping/interleaved capacitor plates (the rotary varactor and the digitally tuned version) — i.e., interpenetrating electrode geometry to increase capacitance per unit area.
• Interdigitated/comb capacitor electrodes were staple MEMS geometry by 2003 (indeed, Dhuler's own background cites an interdigitated capacitive accelerometer). Adding interleaved projections to increase the capacitive-interaction area between the tuning element and fixed electrodes is an obvious, results-predictable adaptation.
• NOTE: the '348 spec itself frames the projections as an optimization ("function to increase the area… which thereby increases the capacitance") — a design optimization, not a patentable advance.

Combination 4 (springs & lifters — claims 18–23, 25, 27): Dhuler '644 + Bauhahn '662 (+ '208)

• Claims 18–23: plural suspension springs with plural lifters acting on each spring; springs routed beneath the electrodes; four springs on four sides/ends; lifter assemblies in which an axially-extending spring runs orthogonally to a curved lifter.
• Bauhahn discloses suspending the movable capacitor element on springs (spiral metallic/dielectric arms) and notes the electrode-to-spring trade-offs.
• Dhuler '644 and the Dhuler arched-beam family (e.g., the thermal arched-beam actuators surfaced in the same search; see the Dhuler inventor listing showing arched-beam/hollow-beam suspension devices) disclose curved/pre-stressed actuator-and-suspension beams — the ancestor of the '348 "lifter."
• Symmetric four-point suspension for pure translational (piston) motion without rotation — claim 22 — is routine MEMS suspension design (the '348 spec presents it as a straightforward stability improvement, FIG. 4B/5B).
• Claims 25 & 27 (pre-stressed beams biasing the tuning element): stress-engineered/curved beams (oxide-encased polysilicon or metal-on-polysilicon bimorphs) are among the oldest MEMS techniques; the '348 spec's own FIG. 4C/4D implementations are standard stress-gradient beams. Combining such a lifter with a suspension spring to set the tuning element's rest elevation is a predictable use of a known actuator.

Combination 5 (posts/anchors — claim 12) and product-by-process (claims 13–14)

• Claim 12 (plural posts anchoring each electrode to the substrate): a trivial, well-known way to rigidly fix a surface-micromachined structure at a set elevation; the '208 platform provides anchors as a matter of course.
• Claims 13–14 ("made by the process of" fabricating the electrodes — and optionally the tuning element — from a common structural layer by surface micromachining): this is a direct application of the '208 five-level process. Defining coplanar electrodes and the tuning element in one polysilicon layer with one mask is the predictable, expressly-beneficial use of that platform (single-mask definition avoids misalignment — the '348 spec's stated rationale). Product-by-process claims are assessed on the structure produced; that structure is fully reachable via '208 + Yoon.

5. Claims I could not read — explicit limitation

The earlier section could not retrieve claims 16, 17, 24, 26, 28, 29, 30, and 31. I therefore do not assert a § 103 mapping for those. Based on the text I do have (every retrieved claim depends from claim 1; the dependent families cluster around conductors, projections, suspensions, lifters, pre-stressed beams), the most likely additional subject matter is further spring/lifter and array detail — but this is inference, not verification. Do not treat any claim 16/17/24/26/28–31 conclusion here as grounded.

6. Graham factor #4 — secondary considerations (no evidentiary support)

• No unexpected results. The '348's asserted benefits (high Q; single-mask coplanar fabrication) are the predictable consequences of choices both Yoon and '208 expressly teach or enable. There is no showing that the reorientation produced an unexpectedly superior result.
• No long-felt-need nexus established. The high-Q MEMS varactor need was recognized throughout the field (Yoon, Dhuler) and was being actively solved by others before the 2003 priority date.
• No commercial-success evidence, and the record is thin: consistent with the earlier sections, the patent was never asserted and never PTAB-tested and reached its anticipated 2024-08-04 term. Nothing in the file supplies secondary-consideration evidence with a nexus to the claims.

7. Flags / contradictions to carry forward

1. New flag (this section): U.S. 6,377,448 (Deane) and U.S. 6,741,449 (Pasternak) — the two references the earlier summary lists as examiner-cited — could not be verified here, and the '449 search returned an unrelated data-transfer-interface '449. This does not contradict the earlier section (which did not rely on them substantively), but the attribution should be confirmed against the '348 front page before being used.
2. Consistency: This analysis is consistent with the earlier "Patent summary" identification of claim 1 as the core independent claim and the dependent families as conductors/projections/springs/lifters/pre-stressed beams. No contradiction with the earlier sections.
3. No date conflict material here; the earlier internal date discrepancy (2026-04-26 vs. 2026-10-01) is irrelevant to § 103.

8. Bottom line

Claims 1–15 (and the dependent spring/lifter/conductor families I could read) are vulnerable to § 103 as a combination of:

• Primary — Yoon, U.S. 6,490,147 (two fixed capacitor plates + a movable tuning element between them to vary capacitance, expressly to avoid suspension-spring series resistance and raise Q, express pull-in and shorting benefits; copper plates for high Q); plus
• the '348 specification's own admissions (movable element inserted between two fixed electrodes; suspension-spring resistance degrading Q); plus
• '208 (the same assignee's five-level surface-micromachining platform) supplying the coplanar, single-structural-layer realization and metallization; with
• Dhuler '644 and Bauhahn '662 as confirming references for the high-Q RF-tunable-capacitor field, MEMS-actuated capacitance variation, interleaved-plate geometry, and conductive-plate/metallization-for-Q teachings.

The single point of departure for claim 1 over Yoon is the coplanar (laterally spaced) electrode orientation with orthogonal tuning-element motion; the remaining elements are either literally in Yoon or admitted prior art. Under KSR, selecting between stacked and coplanar electrodes — a routine design choice dictated by the shared-assignee '208 fabrication platform and yielding a predictable high-Q tunable capacitor — is very likely obvious. The strongest potential defense for the applicant is that no single reference (and, as best I can verify, no cited combination) affirmatively discloses the coplanar fixed-electrode + vertically translating tuning element arrangement, so a defendant would want to locate a reference (or non-patent literature, e.g., lateral/comb-drive tunable capacitors) that does — a search the examiner's cited art, on the record available to me, does not clearly supply.

Caveat: this is a preliminary, best-evidence analysis keyed to the references retrievable on this page. Two cited references are unverified, eight claims are unread, and the precise construction of "first dimension" (whether the tuning element's required motion axis is strictly vertical) controls how close Yoon alone comes. Confirm the full "References Cited" list and the untruncated claims in Patent Center before relying on any ground.