Chemical-Mechanical Planarization (CMP) with Polish-Stop Barrier

This is engineering work-product, not a legal opinion or a clearance/validity search. It names the legal standards; the patentability conclusions are the attorney’s.

How to read this section (the §103 framework, stated once): under Graham/KSR, obviousness turns on the scope and content of the prior art, the differences from the claims, the level of ordinary skill, and objective indicia — and any reason to combine references must be articulated, never hindsight. The per-application notes below POSITION each use against that framework; the deeper, claim-level inventive-step analysis is treated separately, at the level of the mechanism’s claims.

Each application is triaged as either "spec support only" (a predictable same-mechanism use — valuable as disclosure / spec breadth, but unlikely to support its own non-obvious claim) or "claim candidate" (required non-obvious adaptation — may merit a claim of its own).

Integrated-circuit fabrication — device isolation — Removing ridge-like oxide protuberances left by recessed-oxide isolation (and analogous over-fill topography of trench-based isolation) to eliminate conductive rails and step-induced line failures, leaving planar isolation regions.

Closest existing work: Selective oxide-over-nitride STI CMP, where a silicon-nitride layer serves as the polish stop while overfill oxide is planarized down to it (STI CMP review and the ceria/silica high-selectivity SiO2:Si3N4 slurry papers); at the claim level, CN113113304A recites a cap-layer-stopped CMP followed by cap-layer removal.

Differentiation: Abstract-level vs. the STI CMP papers: those use nitride as an underlying mask/stop beneath overfill oxide, whereas this invention's named novel element is a CONFORMAL BLANKET nitride (~100 nm) deposited OVER pre-existing 'bird's head' protuberances so that film-plus-protuberance are abraded only at topographic high points while intact flat nitride self-terminates — coupled with the specific alkaline colloidal-silica slurry (sodium dichloroisocyanurate, Na2CO3·H2O, pH 9.5–12.5) and ~6:1 SiO2:Si3N4 selectivity rather than the ceria chemistries described in the art. §102/§103 for counsel to assess.

Triage: claim candidate — the application required non-obvious adaptation; it may merit a claim of its own

Citations:

• Shallow Trench Isolation Chemical Mechanical Planarization: A Review (2015) — Directly describes colloidal slurries with additives giving high oxide:nitride selectivity while stopping on an underlying nitride film — the closest field-level match to the differential-removal mechanism in this exact domain; compare at abstract level, pull full text. (provenance: grounded)
• Selective Chemical Mechanical Polishing of Silicon Dioxide over Silicon Nitride for Shallow Trench Isolation Using Ceria Slurries (2009) — Selective SiO2-over-Si3N4 CMP for STI, but with ceria (not the claimed alkaline colloidal-silica/sodium dichloroisocyanurate system) and conventional buried-nitride-stop geometry rather than a blanket film over protuberances; abstract-level. (provenance: grounded)
• CN113113304A (verify) — Claim-level: recites CMP stopping on a cap layer followed by removal of the cap layer — structurally analogous stop-then-strip flow, but for a metal feature with a timed second CMP, not a conformal nitride over oxide protuberances with the recited slurry selectivity. (provenance: grounded)

Standard: §102 novelty (for counsel to assess)

Inventive-step consideration (§103 argument): Under §103, the analysis for the isolation application turns on whether the conformal Si3N4 polish-stop plus high-selectivity colloidal-silica slurry (SiO2:Si3N4 ~6:1, self-terminating over flats) would have been obvious over the inventor-cited ROX background (D1–D3) alone or combined with the STI-CMP literature; note the STI-CMP references (2006–2025) recite exactly this stop-on-nitride/selective-oxide-removal mechanism for the same materials, so counsel must first confirm their prior-art status relative to this invention's (apparently earlier ROX-era) priority date, since if they postdate the invention they cannot be combined.

Articulated reason (KSR): The STI-CMP references (esp. the 2015 STI CMP review and the ceria/silica selectivity papers) supply an explicit reason — using a nitride film as a CMP stop with a selective colloidal slurry to planarize isolation oxide is the stated design goal in the same substrate/material set — but the retrieved core-function patents (inkjet printhead, magnetic shift register, NAND) are unrelated art with no articulated reason to combine; against D1–D3 alone, the reason to add a blanket nitride stop and reformulated slurry is the acknowledged failure of prior methods to fully remove bird's-head oxide, though hindsight must be avoided in selecting the specific slurry chemistry.

Reasonable expectation of success: If the STI-CMP literature qualifies as prior art, a PHOSITA would have had a strong expectation of success because those references demonstrate high oxide:nitride selectivity and self-termination on nitride for isolation planarization; against only D1–D3, expectation is weaker — prior methods achieved only partial bird's-head removal, so achieving complete protuberance removal with intact planar nitride via the specific alkaline dichloroisocyanurate/Na2CO3 slurry and ~2.5 psi / 40–50 °C window is not clearly predictable.

Secondary considerations (each needs a nexus):

• Long-felt but unmet need — nexus: Nexus to the self-terminating planar-region protection and complete protuberance removal — D1 (prior polishing gave only partial bird's-head removal) corroborates a recognized deficiency the claimed nitride-stop/selective-slurry combination addresses.
• Teaching away / avoidance of undesirable additives — nexus: The disclosure asserts prior etch-back (D2) required undesirable additives, positioning the specific colloidal-silica/dichloroisocyanurate slurry as an improvement; this rests partly on an asserted characterization of D2 and should be corroborated by the actual D2 text — a reference showing comparable additive-free selectivity would weaken it.
• Unexpected results (specific selectivity/rate window) — nexus: Nexus would attach to the claimed 6:1 optimum selectivity and 12.4–36.5 nm/min nitride rate only if that performance exceeds what the STI-CMP references already report; the L-isoleucine (2025) and ceria-slurry papers describe tunable moderate selectivity, so the asserted advantage over those specific values must be shown rather than assumed — presently uncorroborated as to unexpectedness.
• Commercial success — nexus: No nexus established — no evidence in the record tying market success to the claimed slurry composition or selectivity ratio.

Standard: §103 obviousness (for counsel to assess)

Additive manufacturing — post-processing of printed metal parts — Selectively knocking down localized surface protrusions (spatter/satellite particles, up-skin asperities) on printed components while preserving net-shape geometry, using a self-terminating conformal barrier so critical surfaces are not over-machined.

Closest existing work: Selective STI-type CMP that polishes SiO2 while stopping on a Si3N4 barrier — the same self-terminating polish-stop mechanism, but disclosed only for planar semiconductor wafers, not for finishing 3D printed metal parts. The provided patents (inkjet printhead, magnetic shift register, NAND/CMP semiconductor structures) are unrelated to metal AM finishing.

Differentiation: None of the provided references applies the named blanket conformal polish-stop barrier + chemically-selective self-terminating removal to isolated metal protrusions (spatter/satellites, up-skin asperities) on near-net-shape printed metal parts; the STI CMP art teaches the mechanism only for flat oxide-over-nitride wafers, and says nothing about conformal barrier coating and pressure-concentrated knockdown on non-planar freeform metal geometry (abstract-level comparison — pull full text before relying). §102 novelty and §103 obviousness for counsel to assess.

Triage: claim candidate — the application required non-obvious adaptation; it may merit a claim of its own

Citations:

• Shallow Trench Isolation Chemical Mechanical Planarization: A Review (2015) — Describes the core self-terminating selective-CMP mechanism (oxide removal stopping on nitride via high-selectivity colloidal slurries) that the AM application repurposes; establishes the mechanism is known in the wafer domain but not for 3D metal parts. (provenance: grounded)
• Selective Chemical Mechanical Polishing of Silicon Dioxide over Silicon Nitride for Shallow Trench Isolation Using Ceria Slurries (2009) — Closest to the SiO2:Si3N4 selectivity and nitride polish-stop elements, but confined to planar wafer STI, not conformal-barrier finishing of printed metal protrusions. (provenance: grounded)

Standard: §102 novelty (for counsel to assess)

Inventive-step consideration (§103 argument): Under §103, the retrieved references establish the invention's core mechanism — a conformal polish-stop barrier plus a chemically-selective slurry that removes elevated material and self-terminates on the intact barrier over flats — as well-developed STI CMP art (SON stop-on-nitride, high SiO2:Si3N4 selectivity slurries), so the question for the attorney is whether transplanting that wafer-scale mechanism to knocking down spatter/asperities on 3D printed metal parts is a predictable use of a known technique or requires materials and process choices the references do not supply.

Articulated reason (KSR): The shared 'level only the peaks and stop automatically' function supplies a facial motivation, but every provided reference operates on near-planar silicon wafers with a Si3N4-over-SiO2 chemistry; none addresses metal parts, macroscopic (spatter/satellite) protrusions, or curved net-shape surfaces, so a PHOSITA would have no reference-based teaching identifying an analogous slower-eroding conformal barrier or a metal-selective slurry — reaching the AM metal application from these references risks hindsight rather than an articulated combination.

Reasonable expectation of success: Low as directly transferred: CMP's self-termination depends on a rigid pad contacting a substantially planar wafer so pressure concentrates on sub-micron elevations, whereas AM metal parts present tens-of-micron roughness, millimeter-scale 3D geometry, and metallic (not oxide) protrusions — the named alkaline colloidal-silica/dichloroisocyanurate chemistry and the specific 4:1–40:1 SiO2:Si3N4 (and ~8:1 polysilicon) selectivity ratios have no demonstrated metal analog, so a PHOSITA could not predict that a conformal barrier over a metal protrusion would self-terminate on curved flats.

Secondary considerations (each needs a nexus):

• Long-felt need / recognized finishing bottleneck — nexus: The disclosure asserts conventional AM finishing (machining, tumbling, electropolish) removes material indiscriminately and jeopardizes tolerances, but this rests on an uncorroborated asserted norm — no provided reference documents the AM finishing problem, and the named novel elements (Si3N4 barrier, silica slurry, silicon-specific selectivity ratios) are not shown to be the solution to a metal-part need; a reference establishing metal-selective barrier polishing would defeat any nexus.
• Unexpected results (self-termination on non-planar metal) — nexus: Any unexpected-results argument would need data that the wafer-derived selectivity mechanism functions on 3D metal net-shape surfaces; the AM use is speculative/described-not-demonstrated in the disclosure, so no nexus to the claimed selectivity ratios or slurry composition is established on this record.
• Teaching away — nexus: No provided reference teaches away from applying CMP to metal parts; the STI references are simply silent on metals, which is not a teaching away — so this carries no weight.

Standard: §103 obviousness (for counsel to assess)

Precision / ultra-precision optics fabrication — Removing isolated raised surface defects (fused particles, coating nodules, raised lips of digs) from optical surfaces such as mirrors and laser optics without disturbing the figure of the surrounding surface.

Closest existing work: Selective SiO2-over-Si3N4 CMP with a silicon-nitride polish-stop layer and selectivity-tuned colloidal/ceria slurries, as described in the STI CMP literature — the same topography-selective, stop-on-nitride mechanism the invention uses, but all directed to planarizing patterned semiconductor wafers, not optical surfaces.

Differentiation: At the abstract level (papers give snippets, not claim text; inventor-cited D1–D3 are background only): none of the provided references disclose applying the blanket conformal Si3N4 polish-stop plus alkaline colloidal-silica selective-CMP mechanism to remove isolated raised defects (fused particles, coating nodules, dig lips) from a precision optical figure — they address STI planarization of flat wafers, whereas here the named self-terminating intact-nitride-over-figure element must preserve an optical surface figure rather than a planar active area. §102 novelty and §103 obviousness are for counsel to assess.

Triage: spec support only — a predictable same-mechanism use; valuable as disclosure / spec breadth, but unlikely to support its own non-obvious claim

Citations:

• Shallow Trench Isolation Chemical Mechanical Planarization: A Review (2015) — Describes the core mechanism — colloidal slurries with additives giving high oxide:nitride selectivity while stopping on an underlying Si3N4 film, i.e. the same polish-stop/selective-removal approach applied to STI rather than optics. (provenance: grounded)
• Selective Chemical Mechanical Polishing of Silicon Dioxide over Silicon Nitride for Shallow Trench Isolation Using Ceria Slurries (2009) — Directly on the SiO2:Si3N4 selective CMP with nitride stop, the invention's central selectivity element, but in the semiconductor STI field, not optical-figure preservation. (provenance: grounded)
• CN113113304A (verify) — Claims a CMP process that stops on a cap layer which is later removed — analogous blanket-stop-then-strip sequence, though for metal features in semiconductors, not optics. (provenance: grounded)

Standard: §102 novelty (for counsel to assess)

Condensed entry: triaged spec-support-only, so the full §103 positioning is reserved for claim-candidate applications and the origin field.

Integrated-circuit fabrication — multilevel interconnect / interlayer dielectric — Planarizing dielectric topography over metal features and damascene structures to restore flatness for subsequent lithography and to prevent line breakage over steps, using the intact conformal film to self-terminate and limit dishing/erosion.

Closest existing work: CN113113304A discloses forming a metal feature in a dielectric with a first cap layer, performing CMP that stops on the cap layer, then removing the cap — a stop-layer-controlled CMP for interconnect/dielectric planarization. The STI-CMP body of art (e.g., ceria/silica slurries with SiO2:Si3N4 selectivity stopping on nitride) is the closest to the selectivity/stop-layer mechanism.

Differentiation: Abstract-level (only claim text for CN113113304A is claim-level here): the cited references use a stop/cap layer to halt polishing beneath removed bulk material, whereas the invention's named element is a BLANKET CONFORMAL Si3N4 film (~100 nm) deposited OVER the topography so that the film-plus-protuberance is abraded only where topographically elevated while the intact nitride over flat regions self-terminates — combined with the specific alkaline colloidal-silica slurry (sodium dichloroisocyanurate / Na2CO3·H2O, pH 9.5–12.5) delivering ~6:1 SiO2:Si3N4 selectivity at ~2.5 psi and 40–50 °C; the STI references stop on a patterned nitride over active areas rather than removing a conformal cap only at protrusions. §102 novelty for counsel to assess.

Triage: spec support only — a predictable same-mechanism use; valuable as disclosure / spec breadth, but unlikely to support its own non-obvious claim

Citations:

• CN113113304A (verify) — Cap-layer-controlled CMP of a metal feature in a dielectric with the polish stopping on the cap and subsequent cap removal — closest to the ILD/interconnect planarization use, but does not disclose the blanket conformal over-topography film with self-terminating flat regions or the specific selective slurry. (provenance: grounded)
• US7598097B2 (verify) — Involves forming protuberances and multi-layer stacks with capping layers in dielectric interconnect-like structures, but for magnetic shift registers and via notching — not selective-abrasion protuberance removal by a conformal polish-stop. (provenance: verify_full_doc)

Standard: §102 novelty (for counsel to assess)

Inventive-step consideration (§103 argument): Under §103, this application uses the same conformal-nitride-stop + selective-slurry mechanism claimed for ROX 'bird's head' removal, but now over back-end metal/damascene dielectric topography; the STI-CMP references (2015 Review, 2009 ceria, 2006, 2017 papers) already establish stop-on-Si3N4 selective polishing as the standard endpoint-control approach, so the interconnect use reads as a predictable application of a known mechanism to an adjacent region of the same fabrication flow.

Articulated reason (KSR): The disclosure's own problem statement (metal-line breakage over steps, degraded wiring density) supplies the motivation, and the STI CMP literature supplies the technique — a PHOSITA planarizing interlayer dielectric would apply established selective stop-on-nitride CMP to obtain a self-terminating endpoint that limits dishing/erosion; this is a KSR predictable-use rationale rather than hindsight, though note the invention's blanket conformal nitride over protuberances differs from STI's patterned nitride mask, a distinction not directly supplied by the STI art.

Reasonable expectation of success: High for the SiO2-over-SiO2 interlayer-dielectric case because the demonstrated SiO2:Si3N4 selectivity (~6:1) and alkaline colloidal-silica slurry chemistry transfer directly to oxide ILD; expectation is weaker where the interconnect involves low-k dielectrics or exposed metal, since selectivity and slurry compatibility for those materials are neither demonstrated nor shown in the cited references.

Secondary considerations (each needs a nexus):

• Unexpected results / self-terminating dishing-erosion control — nexus: Nexus to the intact-nitride-over-flat-regions feature, but this application is described_not_demonstrated (no interconnect data), so any unexpected-result argument for the ILD use is presently unsupported by the disclosure's own data.
• Long-felt need (line breakage over steps, wiring-density degradation) — nexus: Nexus to the planarization function is asserted in the problem statement, but the need for planarization in multilevel interconnect is broadly known and CMP is the established solution, so this carries limited weight absent evidence the specific blanket-nitride-stop approach solved a need others failed to meet.
• Teaching away / asserted norm that CMP over-removes flat regions without a stop — nexus: Rests on an uncorroborated asserted norm; the provided STI-CMP references already teach nitride stop layers and high-selectivity slurries to limit nitride loss and dishing, which tends to defeat rather than support any teaching-away for the interconnect extension.

Standard: §103 obviousness (for counsel to assess)

Data-storage device fabrication — thin-film magnetic recording heads — Planarizing head stacks and controlling pole-tip recession/protrusion at interfaces between dissimilar hard and soft materials, using a conformal slow-eroding stop that self-terminates over co-planar regions.

Closest existing work: STI CMP practice where a silicon-nitride layer serves as a polish stop and selective oxide-over-nitride slurries level topography (the retrieved STI review and selective SiO2/Si3N4 slurry papers). None of the retrieved references address planarizing thin-film magnetic recording head stacks or pole-tip recession/protrusion at hard/soft material interfaces.

Differentiation: Abstract-level (retrieved items are papers/snippets or unrelated patent claims): the invention's named elements — a CONFORMAL blanket Si3N4 polish-stop barrier deposited over pre-existing protuberances that self-terminates removal over co-planar regions, with the specified alkaline colloidal-silica slurry and ~6:1 SiO2:Si3N4 selectivity — differ from STI CMP where nitride is a patterned active-area mask (not a conformal blanket over topography) and from the retrieved patents (inkjet, shift register, NAND, timed-CMP cap removal) which do not use a conformal self-terminating stop over dissimilar magnetic-head materials. §102/§103 for counsel to assess.

Triage: spec support only — a predictable same-mechanism use; valuable as disclosure / spec breadth, but unlikely to support its own non-obvious claim

Citations:

• Paper: Shallow Trench Isolation CMP: A Review (2015) — Same mechanism — colloidal slurry with additives giving high oxide:nitride selectivity while stopping on nitride; closest core-function analog but for STI, not magnetic heads or conformal-over-protuberance leveling. (provenance: grounded)
• Paper: Selective CMP of SiO2 over Si3N4 for STI Using Ceria Slurries (2009) — Selective oxide-over-nitride CMP with a nitride stop layer — same differential-selectivity principle the invention relies on. (provenance: grounded)
• CN113113304A (verify) — CMP that stops on a cap layer and then removes the cap — conceptually related stop-layer/self-terminating idea, but timed two-speed metal CMP, not a conformal self-terminating barrier over protuberances. (provenance: verify_full_doc)

Standard: §102 novelty (for counsel to assess)

Condensed entry: triaged spec-support-only, so the full §103 positioning is reserved for claim-candidate applications and the origin field.

Precision manufacturing — chemically-assisted finishing fluid (composition repurposing) — Using the disclosed oxidizing, alkaline, buffered abrasive medium as a stand-alone lapping/finishing fluid for hard engineering metals and ceramics (bearings, seals, medical and aerospace components), independent of the barrier-based planarization method.

Closest existing work: Colloidal-silica STI CMP slurries described in the STI CMP review literature and selective-silica-abrasive papers — alkaline colloidal-silica suspensions engineered for oxide/nitride selectivity in semiconductor planarization. None address use as a stand-alone lapping/finishing fluid for hard engineering metals or ceramics.

Differentiation: Abstract-level (no claim text on the slurry chemistry in the retrieved art): the prior art teaches colloidal-silica slurries for SiO2/Si3N4 selectivity in CMP, but none disclose the specific oxidizing-plus-abrasive medium (sodium dichloroisocyanurate oxidizer + Na2CO3·H2O buffer + colloidal silica, pH 9.5–12.5) repurposed as a surface-conversion/abrasion finishing fluid for hard metals and ceramics; the named oxidizer-driven surface-conversion chemistry for metal/ceramic substrates is absent from the CMP-selectivity references. §102 novelty and §103 obviousness for counsel to assess.

Triage: spec support only — a predictable same-mechanism use; valuable as disclosure / spec breadth, but unlikely to support its own non-obvious claim

Citations:

• Shallow Trench Isolation Chemical Mechanical Planarization: A Review (2015) — Surveys alkaline colloidal-silica slurries with additives for oxide/nitride CMP — closest disclosure of the base composition, but confined to semiconductor planarization, not standalone metal/ceramic finishing (provenance: grounded)
• Highly selective chemical mechanical polishing of Si3N4 over SiO2 using advanced silica abrasive (2017) — Describes advanced silica-abrasive slurry chemistry for CMP selectivity; relevant to the abrasive-suspension composition but not its repurposing as a hard-substrate finishing fluid (provenance: grounded)

Standard: §102 novelty (for counsel to assess)

Condensed entry: triaged spec-support-only, so the full §103 positioning is reserved for claim-candidate applications and the origin field.

Of 6 application(s) analyzed for prior art:

• 6 grounded (claim text was available for 5 of 14 retrieved references; the comparison is abstract-level for the rest)