1014-00004 · Hawthorne Vega & Cruz LLP · Infringement Contentions & Claim Charts

doe-analysis-memo.docx DOCUMENT Litigation

103 / 168 ← → ✕

PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT

MEMORANDUM

PRIVILEGED AND CONFIDENTIAL — ATTORNEY-CLIENT PRIVILEGE / ATTORNEY WORK PRODUCT DOCTRINE

This memorandum was prepared by attorneys at Calderwood & Harkness LLP at the direction of, and for the benefit of, our client Optiwave Semiconductor Corp. ("OWS" or "Client") in connection with pending litigation and is protected by the attorney-client privilege and the attorney work product doctrine. This memorandum contains mental impressions, conclusions, opinions, and legal theories of the attorneys involved in this litigation. It is not intended for disclosure to any third party and should not be disclosed without the express written consent of the undersigned.

Right-click to update Table of Contents

I. Executive Summary

This memorandum analyzes the viability of asserting infringement under the doctrine of equivalents ("DOE") for three specific patent claims asserted by Optiwave Semiconductor Corp. ("OWS") against Lumicore Technologies, Inc. ("Lumicore") in the above-captioned matter: Claims 3 and 7 of U.S. Patent No. 10,891,193 (the "'193 Patent") and Claim 12 of U.S. Patent No. 11,402,015 (the "'015 Patent"). This analysis is undertaken in advance of OWS's obligation to serve its P.R. 3-1 Infringement Contentions on September 16, 2024, pursuant to the Court's scheduling order, and is intended to inform the litigation team's decision-making regarding which DOE theories to include in those contentions.

OWS accuses three Lumicore product families of infringing the asserted claims: the LumiLink 400G-ZR+ (accused of infringing all three claims), the LumiLink 800G-DR8 (accused of infringing all three claims), and the LumiEdge 1600G-LR (accused of infringing '193 Patent Claims 3 and 7 and '015 Patent Claim 12). These products represent a combined FY 2023 revenue of approximately $264.9 million and are central to Lumicore's silicon photonics transceiver product line.

The DOE analysis in this memorandum focuses on two specific claim limitations where literal infringement may be contested: (1) the "coherent detection module" limitation in '193 Patent Claims 3 and 7, and (2) the "two-dimensional sub-wavelength grating" limitation in '015 Patent Claim 12. In both instances, Lumicore's accused products employ structures that depart from the literal claim language as narrowed during prosecution but that perform substantially the same function, in substantially the same way, to achieve substantially the same result as the claimed elements.

Bottom-Line Assessment. DOE contentions are supportable for all three claims. However, prosecution history estoppel presents a moderate risk with respect to '193 Patent Claims 3 and 7 and a lower-but-non-trivial risk with respect to '015 Patent Claim 12. In both cases, the claim limitations at issue were narrowed during prosecution to overcome prior-art rejections, triggering the Festo presumption of surrender. We believe this presumption is rebuttable under the "tangential relation" exception recognized by the Supreme Court in Festo Corp. v. Fenmore & Co., 535 U.S. 722 (2002), and we recommend proceeding with that exception as the primary argument for overcoming estoppel on both patents. The unforeseeability exception provides a secondary basis, particularly for '015 Patent Claim 12.

We recommend proceeding with DOE contentions on all three claims in the P.R. 3-1 Infringement Contentions. This belt-and-suspenders approach preserves the maximum scope of OWS's infringement theories, provides strategic flexibility through Markman and dispositive motions, and ensures that OWS's damages case captures the full scope of Lumicore's infringing activity. The detailed recommendations are set forth in Section VI of this memorandum.

II. Background

A. The Asserted Patents and Claims

1. The '193 Patent — U.S. Patent No. 10,891,193. The '193 Patent, titled "High-Bandwidth Coherent Optical Transceiver Integrated on Silicon Substrate," was filed on April 7, 2017 and issued on January 12, 2021. The original assignee of record is Optiwave Semiconductor Corp. The '193 Patent claims the benefit of U.S. Provisional Application No. 62/499,117, filed April 7, 2017, which is a common provisional application shared with the '015 Patent discussed below. The '193 Patent was prosecuted by Dr. Martin Chu of Cascade IP Advisors, LLC (1550 Crystal Drive, Suite 600, Arlington, VA 22202), who served as OWS's primary patent prosecution counsel during the relevant period. The sole named inventor is Dr. Jian-Wei Huang, OWS's Chief Technology Officer, who conceived of the invention during OWS's development of its next-generation silicon photonic transceiver platform.

The '193 Patent is directed generally to an integrated coherent optical transceiver fabricated on a silicon substrate, enabling high-bandwidth data transmission and reception using coherent detection techniques. The patent specification describes, in considerable detail, the architecture of the transceiver, including the integration of a transmitter path and a receiver path on a common silicon-on-insulator (SOI) substrate. A central feature of the receiver path is the "coherent detection module," which the specification describes as a key structural component responsible for mixing received optical signals with a local oscillator (LO) to extract both phase and amplitude information from the incoming signal.

The claims at issue for DOE analysis in this memorandum are Claim 3 (which depends directly from independent Claim 1) and Claim 7 (which depends from Claim 1 through intervening Claim 5). Both Claims 3 and 7 recite the "coherent detection module" limitation, which is the focus of the DOE analysis. As discussed in detail in Section III below, Lumicore's accused products may not literally satisfy this limitation because they employ a modified coherent detection architecture that departs from the specific structural and functional features recited in the amended claim language. However, the accused architecture performs the same coherent detection function by the same fundamental operating principle, achieving the same result, making it a strong candidate for DOE analysis.

2. The '015 Patent — U.S. Patent No. 11,402,015. The '015 Patent, titled "Polarization-Diverse Grating Coupler for Silicon Photonic Transceivers," was filed on August 30, 2019 and issued on August 2, 2022. Like the '193 Patent, the '015 Patent was originally assigned to Optiwave Semiconductor Corp. and claims the benefit of the same U.S. Provisional Application No. 62/499,117, filed April 7, 2017. The '015 Patent was also prosecuted by Dr. Martin Chu at Cascade IP Advisors, LLC.

The '015 Patent is directed to a polarization-diverse grating coupler designed for use in silicon photonic transceivers. The specification describes the challenges of coupling light from a standard optical fiber into a silicon photonic integrated circuit (PIC) while maintaining sensitivity to both transverse-electric (TE) and transverse-magnetic (TM) polarization modes. The patent teaches the use of a two-dimensional sub-wavelength grating structure that simultaneously couples both polarization states from the fiber into the waveguide layer, splitting them into separate waveguide paths for independent processing. This approach is described as superior to prior-art solutions that required separate grating couplers or polarization-diversity schemes external to the coupling element.

The claim at issue for DOE analysis is Claim 12, which is a dependent claim. The key limitation subject to DOE analysis is the "two-dimensional sub-wavelength grating" limitation. As discussed in Section IV below, Lumicore's accused products employ a grating structure that achieves the same polarization-diverse coupling function but may depart from the literal "two-dimensional sub-wavelength" geometry specified in the amended claim language. The accused grating structure achieves its polarization-diversity function through a structurally analogous but not literally identical grating geometry, making it a viable candidate for DOE.

B. Accused Products

OWS accuses three Lumicore product families of infringing the asserted claims. Each of these products is a pluggable optical transceiver module designed for high-speed data center interconnect applications and is manufactured using Lumicore's proprietary silicon photonic platform. The relevant products and their commercial significance are as follows:

LumiLink 400G-ZR+. Lumicore launched the LumiLink 400G-ZR+ in the third quarter of 2022 as a coherent pluggable transceiver module designed for long-haul and metro data center interconnect applications. The 400G-ZR+ implements a coherent transceiver architecture on a silicon photonic die, incorporating both a coherent transmitter and a coherent receiver in a single QSFP-DD form factor. Lumicore reported FY 2023 revenue of approximately $94.8 million for the 400G-ZR+ product line. The LumiLink 400G-ZR+ is accused of infringing all three claims at issue ('193 Patent Claims 3 and 7, and '015 Patent Claim 12).

LumiLink 800G-DR8. Launched in the first quarter of 2023, the LumiLink 800G-DR8 is a short-reach (500m) transceiver module designed for intra-data-center applications at 800 Gbps aggregate bandwidth. The 800G-DR8 employs eight lanes of 100G PAM4 signaling on a silicon photonic platform, with a polarization-diverse grating coupler architecture for fiber coupling. Lumicore reported FY 2023 revenue of approximately $128.4 million for the 800G-DR8 product line, making it Lumicore's highest-revenue accused product. The LumiLink 800G-DR8 is accused of infringing all three claims at issue.

LumiEdge 1600G-LR. Lumicore launched the LumiEdge 1600G-LR in the fourth quarter of 2023 as its next-generation long-reach transceiver module, targeting 1.6 Tbps data rates for next-generation data center interconnect. The 1600G-LR is the newest of the accused products and represents Lumicore's entry into the emerging 1.6T market segment. Lumicore reported FY 2023 revenue of approximately $41.7 million for the 1600G-LR product line (reflecting only one quarter of commercial availability). The LumiEdge 1600G-LR is accused of infringing '193 Patent Claims 3 and 7 and '015 Patent Claim 12.

The total accused product revenue for FY 2023 is $264.9 million across all three product families. This figure is expected to increase substantially in FY 2024 as the 800G-DR8 and 1600G-LR products gain broader market adoption.

Benchmark Semiconductor Analysis, LLC ("Benchmark") was retained to perform physical teardowns and structural analysis of all three accused products. Kevin Ng, Lead Engineer at Benchmark, supervised all teardown activities. The initial teardown report for the LumiLink 400G-ZR+ was delivered on December 12, 2022. Follow-on teardown reports were completed for the LumiLink 800G-DR8 (delivered June 2023) and the LumiEdge 1600G-LR (delivered February 2024). The Benchmark teardown data, combined with Lumicore design documents produced in discovery, form the factual foundation for the DOE analysis in this memorandum. Benchmark's findings on the relevant structural features are summarized in Appendix C.

C. Legal Standard for the Doctrine of Equivalents

The doctrine of equivalents is a judicially created doctrine that allows a patentee to claim infringement even when the accused product does not literally satisfy every limitation of a patent claim. The Supreme Court reaffirmed the continuing vitality of the doctrine in Warner-Jenkinson Co. v. Hilton Davis Chemical Co., 520 U.S. 17 (1997), holding that the doctrine of equivalents remains an important part of patent law and is applied on an element-by-element basis.

Under the All-Elements Rule, the doctrine of equivalents must be applied to individual claim limitations, not to the claim as a whole. Warner-Jenkinson, 520 U.S. at 29. A patentee must demonstrate that each element of the claim is met in the accused product, either literally or by an equivalent. The doctrine cannot be applied in a manner that effectively vitiates (reads out entirely) a claim limitation. Id. at 39 n.8.

Courts have recognized two formulations for determining whether an element of an accused product is equivalent to a claim limitation:

1. The Function-Way-Result Test (derived from Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 608 (1950)): Under this test, an accused element is equivalent to a claim limitation if it performs substantially the same function, in substantially the same way, to achieve substantially the same result as the claimed element.

2. The Insubstantial Differences Test: Under this alternative formulation, infringement under the doctrine of equivalents is found when the differences between the claim limitation and the accused element are insubstantial. See Warner-Jenkinson, 520 U.S. at 40.

This memorandum primarily applies the function-way-result test, as is standard in Eastern District of Texas practice and consistent with Judge Gilstrap's prior rulings in patent cases. See, e.g., Omega Patents, LLC v. CalAmp Corp., No. 2:13-cv-00726-JRG, 2017 WL 3425704 (E.D. Tex. Aug. 9, 2017) (Gilstrap, J.) (applying function-way-result test in evaluating DOE contentions at summary judgment). The insubstantial-differences test may provide additional support but is not the primary framework for analysis.

Critically, equivalence is assessed at the time of infringement, not at the time the patent was issued. Warner-Jenkinson, 520 U.S. at 37. This temporal requirement means that technologies developed after the patent's issuance can still be captured by the doctrine of equivalents if they are equivalent to the claimed element at the time of the alleged infringement. This principle may be relevant here, particularly with respect to Lumicore's design techniques that were implemented in products launched in 2022 and 2023.

D. Legal Standard for Prosecution History Estoppel

The doctrine of prosecution history estoppel operates as a limitation on the doctrine of equivalents. Where a patentee has narrowed the scope of a claim during prosecution for reasons related to patentability, the patentee may be estopped from later recapturing through the doctrine of equivalents the subject matter that was surrendered by the narrowing amendment.

The Supreme Court set forth the governing framework for prosecution history estoppel in Festo Corp. v. Fenmore & Co., 535 U.S. 722 (2002). Under Festo, when a claim limitation is narrowed by amendment during prosecution for a reason related to patentability — such as to overcome a prior-art rejection under 35 U.S.C. § 102 or § 103 — a presumption arises that the patentee has surrendered all territory between the original claim limitation and the amended claim limitation. Festo, 535 U.S. at 740. This presumption applies regardless of whether the amendment was made at the suggestion of the examiner or was voluntarily offered by the applicant; what matters is whether the amendment was made for a reason related to patentability. Id. at 736–37.

The Festo presumption of surrender is, however, rebuttable. The Supreme Court recognized three circumstances under which a patentee may overcome the presumption:

1. The Unforeseeability Exception: The equivalent at issue was unforeseeable at the time of the narrowing amendment. Festo, 535 U.S. at 740–41. If the alleged equivalent was based on technology that did not exist or was not known in the relevant art at the time of the amendment, the patentee could not reasonably have been expected to have claimed it, and the presumption of surrender does not apply.

2. The Tangential Relation Exception: The rationale underlying the narrowing amendment bore only a tangential relation to the equivalent in question. Id. at 740–41. If the reason the applicant narrowed the claim had nothing (or only a peripheral connection) to do with the specific equivalent now at issue, the patentee should not be foreclosed from asserting equivalence with respect to that particular variation.

3. The "Other Reason" Exception: There is some other reason suggesting that the patentee could not reasonably have been expected to have described the insubstantial substitute in question. Id. This exception serves as a catch-all but is narrowly applied by the Federal Circuit. See Festo Corp. v. Fenmore & Co., 344 F.3d 1359, 1369 (Fed. Cir. 2003) (en banc) (on remand).

The burden of overcoming the Festo presumption rests squarely on the patentee. Festo, 535 U.S. at 740. This is a meaningful burden, and the patentee must present specific, persuasive evidence and argument to rebut the presumption. The Federal Circuit has emphasized that "[t]he patentee bears the burden of showing that the amendment does not surrender the particular equivalent in question." Honeywell Int'l, Inc. v. Hamilton Sundstrand Corp., 370 F.3d 1131, 1142 (Fed. Cir. 2004) (en banc).

It is also important to note that the Festo framework applies to voluntary amendments, not just those made in direct response to examiner rejections. Festo, 535 U.S. at 736–37. The inquiry is whether the amendment was made "for a substantial reason related to patentability," which includes amendments offered proactively by the applicant to advance prosecution, so long as they narrow the claim scope and relate to patentability considerations.

III. Analysis of '193 Patent — Claims 3 and 7

A. Claim Language

The relevant claims of the '193 Patent are set forth in full in Appendix D. For purposes of this analysis, we reproduce the pertinent claim language below.

Claim 1 (independent) recites:

1. An integrated coherent optical transceiver, comprising:

a silicon-on-insulator substrate;

a transmitter path integrated on the silicon-on-insulator substrate, the transmitter path comprising a laser source optically coupled to a Mach-Zehnder modulator array configured to modulate an optical carrier signal with a data signal;

a receiver path integrated on the silicon-on-insulator substrate, the receiver path comprising:

an input waveguide configured to receive a modulated optical signal from an optical fiber;

a local oscillator source integrated on the silicon-on-insulator substrate; and

a signal processing unit configured to recover a data signal from the modulated optical signal;

wherein the transmitter path and the receiver path are monolithically integrated on the silicon-on-insulator substrate and share at least one common optical component.

Claim 3 (dependent on Claim 1) recites:

3. The integrated coherent optical transceiver of claim 1, wherein the receiver path further comprises a coherent detection module disposed between the input waveguide and the signal processing unit, the coherent detection module comprising:

an optical hybrid configured to mix the modulated optical signal with a local oscillator signal from the local oscillator source to produce at least four mixed optical signals;

a plurality of balanced photodetectors, each balanced photodetector configured to receive a respective pair of the mixed optical signals and to produce a differential electrical signal proportional to an optical phase difference; and

a transimpedance amplifier array electrically coupled to the plurality of balanced photodetectors,

wherein the optical hybrid, the plurality of balanced photodetectors, and the transimpedance amplifier array are integrated on the silicon-on-insulator substrate as a unitary coherent detection module.

Claim 7 (dependent on Claim 1 through Claim 5) recites:

7. The integrated coherent optical transceiver of claim 5, wherein the coherent detection module is configured to detect both in-phase (I) and quadrature (Q) components of the modulated optical signal by producing at least eight mixed optical signals from the optical hybrid, and wherein the coherent detection module further comprises a polarization beam splitter integrated upstream of the optical hybrid to separate orthogonal polarization states of the modulated optical signal into two polarization paths, each polarization path directed to a respective optical hybrid.

The "coherent detection module" limitation, which is the focus of this DOE analysis, appears in both Claims 3 and 7. The specification of the '193 Patent describes the coherent detection module in detail at column 7, lines 12–55, and column 8, lines 1–34. The specification states that "[t]he coherent detection module 300 is a key component of the receiver path, enabling the extraction of both phase and amplitude information from the incoming optical signal through coherent mixing with a local oscillator." '193 Patent, col. 7, ll. 12–16. The specification further describes the specific structural arrangement of the optical hybrid, balanced photodetectors, and transimpedance amplifier array as a "unitary" module integrated on the SOI substrate. Id., col. 7, ll. 23–42.

B. Prosecution History of the '193 Patent — The "coherent Detection Module" Amendment

The prosecution history of the '193 Patent is directly relevant to the DOE analysis because the "coherent detection module" limitation was narrowed during prosecution in a manner that triggers the Festo presumption of prosecution history estoppel. The relevant prosecution history excerpts are included in Appendix A; we summarize the key events below.

On March 18, 2019, the U.S. Patent and Trademark Office ("PTO") issued a non-final Office Action rejecting original Claim 1 and all dependent claims (including the predecessors to current Claims 3 and 7) under 35 U.S.C. § 102(a)(1) based on Takahashi et al., "Dense WDM Silicon Photonic MUX with Integrated Coherent Receiver," IEEE Photonics Technology Letters, Vol. 26, No. 6, pp. 600–603 (March 15, 2014) ("Takahashi"). The examiner found that Takahashi disclosed an integrated silicon photonic transceiver with a coherent receiver module that anticipated the originally filed claims.

In the original claim language, the "coherent detection module" limitation was drafted more broadly. The original claim language did not include the requirement that the optical hybrid, balanced photodetectors, and transimpedance amplifier array be integrated on the SOI substrate "as a unitary coherent detection module," nor did it specify the structural arrangement of the optical hybrid to produce "at least four mixed optical signals." The original language encompassed any coherent detection approach integrated on a silicon substrate, without the structural and functional specificity that appears in the issued claims.

On September 16, 2019, the applicant (through Dr. Martin Chu at Cascade IP Advisors) filed a response to the Office Action that included amendments to the pending claims. Specifically, the response amended the "coherent detection module" limitation to add the structural and functional requirements present in the issued claim language — i.e., requiring the specific configuration of the optical hybrid producing at least four mixed optical signals, the balanced photodetectors receiving respective pairs of mixed optical signals, the transimpedance amplifier array, and the integration of these components as a "unitary coherent detection module" on the SOI substrate. In the accompanying remarks, Dr. Chu argued that the amended claim language distinguished over Takahashi because Takahashi's coherent receiver lacked the specific structural integration and unitary module configuration recited in the amended claims. The remarks stated, in relevant part:

"Amended Claim 1 and its dependent claims are patentably distinct from Takahashi. Takahashi describes a coherent receiver section within a WDM multiplexer/demultiplexer structure, but does not teach or suggest the specific unitary coherent detection module configuration recited in the amended claims, in which the optical hybrid, balanced photodetectors, and transimpedance amplifier array are integrated on the SOI substrate as a single, self-contained detection module. Takahashi's receiver architecture distributes these functions across separate, non-unitary components that are not co-integrated in the manner claimed."

The examiner subsequently allowed the claims. A Notice of Allowance was issued on July 8, 2020, and the patent issued on January 12, 2021.

The critical point for DOE analysis is that the narrowing amendment to the "coherent detection module" limitation was made expressly to overcome the § 102 rejection based on Takahashi — that is, for a reason related to patentability. This triggers the Festo presumption that the patentee surrendered the territory between the original broader claim language and the narrowed "coherent detection module" limitation as amended. The scope of surrender encompasses coherent detection approaches that would have been covered by the original broader claim language but that fall outside the narrowed limitation specifying the unitary module configuration with the particular structural requirements.

C. Lumicore's Accused Design — How It Differs from the Literal Claim

Based on the Benchmark teardown data and Lumicore's design documents produced in discovery (Bates Nos. LUMI-00042187 through LUMI-00042643 and LUMI-00078219 through LUMI-00078456), the accused products — the LumiLink 400G-ZR+, LumiLink 800G-DR8, and LumiEdge 1600G-LR — each employ a coherent detection architecture on a silicon photonic platform. The coherent detection functionality in the accused products accomplishes the same overarching purpose as the "coherent detection module" recited in Claims 3 and 7: extracting phase and amplitude information from an incoming optical signal through coherent mixing with a local oscillator.

However, the accused products' coherent detection architecture differs from the literal claim language in a critical structural respect. Specifically, rather than employing a "unitary coherent detection module" in which the optical hybrid, balanced photodetectors, and transimpedance amplifier array are integrated as a single, self-contained module on the SOI substrate (as required by the amended claim language), Lumicore's accused products employ a distributed coherent detection architecture. In this architecture, the optical hybrid function is performed by a multimode interference (MMI) coupler network that is physically separated from the photodetector array on the die, with the two functional blocks interconnected through a set of intermediate silicon waveguides. The transimpedance amplifier (TIA) array in the accused products is implemented on a separate electronic die that is flip-chip bonded onto the silicon photonic die, rather than being monolithically integrated on the SOI substrate as part of a unitary module.

Kevin Ng's Benchmark teardown report for the LumiLink 400G-ZR+ (December 12, 2022) specifically noted that "[t]he coherent receiver section of the LumiLink 400G-ZR+ employs a distributed architecture in which the optical mixing and photodetection functions are spatially separated on the PIC, with the TIA implemented on a co-packaged CMOS die." Benchmark Teardown Report at 37. Similar observations were made in the follow-on teardown reports for the LumiLink 800G-DR8 (June 2023 report at 42) and the LumiEdge 1600G-LR (February 2024 report at 29).

The specific structural difference — the distributed (rather than unitary) module architecture and the separate TIA die — means that Lumicore's accused products may not literally satisfy the "coherent detection module" limitation as narrowed during prosecution. Lumicore's defense counsel will undoubtedly argue that the accused products lack a "unitary coherent detection module" as required by the claim and, therefore, do not literally infringe. This is the precise gap that the doctrine of equivalents is designed to bridge.

D. Function-Way-Result Analysis

We apply the tripartite function-way-result test to the "coherent detection module" limitation of Claims 3 and 7 as applied to Lumicore's accused products.

1. Function. The claimed "coherent detection module" functions to detect modulated optical signals using coherent detection techniques — specifically, by mixing the incoming modulated optical signal with a local oscillator signal to extract both phase and amplitude information from the received signal. This function is fundamental to coherent optical communication and enables the recovery of complex modulation formats (such as QPSK, 16-QAM, and higher-order formats) that carry information in both the amplitude and phase of the optical carrier.

Lumicore's accused coherent detection architecture performs substantially the same function. The distributed MMI coupler network in the accused products mixes the incoming modulated optical signal with a local oscillator signal to produce mixed optical signals carrying phase information. The balanced photodetectors in the accused products detect these mixed signals and produce differential electrical signals proportional to optical phase differences. The co-packaged TIA array amplifies the resulting electrical signals. In all material respects, the function of the accused architecture is identical to the function of the claimed coherent detection module: coherent detection of optical signals to extract phase and amplitude information.

There is no meaningful dispute regarding function. Both the claimed element and the accused element are coherent detection systems that extract phase and amplitude information from incoming modulated optical signals through coherent mixing with a local oscillator. The accused products' promotional materials, including Lumicore's own product datasheets (Bates No. LUMI-00015832), describe the products as implementing "coherent detection" on a "silicon photonic platform" — language that directly aligns with the function of the claimed limitation.

Conclusion: The accused element performs substantially the same function as the claimed "coherent detection module."

2. Way. The more nuanced question is whether the accused design operates in "substantially the same way" as the claimed coherent detection module. The accused products employ a distributed architecture rather than a unitary module, and the TIA is implemented on a separate electronic die rather than being monolithically integrated on the SOI substrate. Lumicore will argue that these structural differences represent a fundamentally different "way" of achieving coherent detection.

However, the operational principle of the accused architecture is the same as the claimed element. Both the claimed and accused systems rely on the same fundamental mechanism: coherent mixing of the incoming signal light with local oscillator light using an optical hybrid structure (whether implemented as a 90° hybrid or an MMI coupler network), followed by detection of the resulting interference signals using balanced photodetectors, and amplification by transimpedance amplifiers. The physics of the coherent detection process — the interference of signal and LO fields on the photodetector surfaces, the balanced detection to suppress common-mode noise, and the transimpedance amplification of the photocurrent — is identical in both the claimed and accused designs.

The structural difference between a "unitary" module and a "distributed" architecture is, in our analysis, a difference of physical layout and packaging, not a difference in the fundamental mechanism of operation. The MMI coupler network in the accused products performs the same optical mixing function as the optical hybrid in the claimed limitation, using the same physical principle of multimode interference to generate the required phase-shifted optical outputs. The balanced photodetectors in the accused products are germanium-on-silicon (Ge-on-Si) photodiodes — the same photodetector technology described in the '193 Patent specification. The only structural difference is that these components are distributed across a larger area of the PIC and interconnected by waveguides rather than being co-located in a single, compact "unitary" module.

With respect to the TIA placement, while the accused products implement the TIA on a separate CMOS die that is flip-chip bonded to the PIC, the TIA still functions as the final amplification stage of the coherent detection chain. The use of co-packaged or flip-chip-bonded electronic dies for TIA implementation is a design choice driven by the practical difficulty of fabricating high-performance TIAs in a silicon photonic process, and does not alter the fundamental "way" in which coherent detection is achieved.

Benchmark's teardown data supports this analysis. Kevin Ng's reports note that the optical mixing, photodetection, and amplification functions in the accused products operate on the same physical principles as described in the '193 Patent, with the structural differences limited to the spatial distribution and packaging of the functional elements. See Benchmark 400G-ZR+ Teardown Report at 38–39; Benchmark 800G-DR8 Teardown Report at 43–44.

We anticipate that Lumicore will argue that the distributed architecture represents a different "way" because the spatial separation of the MMI coupler network and the photodetector array introduces different optical path lengths, potentially different phase-matching conditions, and different noise characteristics compared to a unitary module. This counterargument has some surface appeal but is ultimately unpersuasive. The differences in optical path length and phase matching are compensated for by Lumicore's waveguide routing design (as documented in Lumicore's own design files, Bates No. LUMI-00042301 at 12), and the noise characteristics of the balanced detection scheme remain fundamentally the same. The accused design does not employ a qualitatively different detection mechanism; it simply distributes the same functional elements across a different physical layout.

Conclusion: The accused element operates in substantially the same way as the claimed "coherent detection module." The differences in physical layout and TIA packaging are differences of degree, not of kind, and do not alter the fundamental operational mechanism of coherent optical detection.

3. Result. Both the claimed "coherent detection module" and the accused coherent detection architecture achieve substantially the same result: high-bandwidth coherent detection of modulated optical signals, with recovery of both in-phase and quadrature components of the signal, enabling the demodulation of complex modulation formats on a silicon-integrated photonic platform.

The performance specifications of the accused products confirm that they achieve the same end result as the claimed invention. The LumiLink 400G-ZR+ achieves 400 Gbps coherent transmission and reception; the LumiLink 800G-DR8 achieves 800 Gbps aggregate bandwidth using coherent-like detection in combination with PAM4 signaling; and the LumiEdge 1600G-LR achieves 1.6 Tbps aggregate bandwidth. In each case, the coherent detection architecture in the accused products recovers phase and amplitude information from the received optical signal, enabling high-spectral-efficiency communication — precisely the result described in the '193 Patent specification for the claimed coherent detection module.

Lumicore's product datasheets (Bates No. LUMI-00015832, LUMI-00015847, LUMI-00015863) describe the products as achieving "coherent detection with full phase recovery" on a "silicon photonic platform," language that directly mirrors the result described in the '193 Patent.

Conclusion: The accused element achieves substantially the same result as the claimed "coherent detection module."

4. Overall DOE Assessment for '193 Claims 3 and 7. Based on the foregoing function-way-result analysis, we conclude that the accused products — the LumiLink 400G-ZR+, LumiLink 800G-DR8, and LumiEdge 1600G-LR — infringe Claims 3 and 7 of the '193 Patent under the doctrine of equivalents with respect to the "coherent detection module" limitation. The accused distributed coherent detection architecture performs substantially the same function (coherent detection to extract phase and amplitude information), in substantially the same way (coherent mixing of signal and LO light using an optical hybrid/MMI coupler, balanced photodetection, and transimpedance amplification), to achieve substantially the same result (high-bandwidth coherent optical signal recovery on a silicon photonic platform) as the claimed unitary coherent detection module.

Dr. Caroline Ashford of the Northfield Institute of Technology, OWS's retained technical expert, is expected to support this analysis in her infringement expert report, which is due on June 15, 2025 under the Court's scheduling order. Dr. Ashford has reviewed the Benchmark teardown data and preliminary Lumicore design documents and has indicated in preliminary discussions that the function-way-result analysis is well-supported by the technical evidence. Her expert report should present this analysis in detail, with reference to specific teardown findings, Lumicore design files, and published technical literature on coherent detection architectures.

E. Prosecution History Estoppel Risk for '193 Claims 3 and 7

1. The Festo Presumption Applies. As discussed in Section III.B above, the "coherent detection module" limitation in Claims 3 and 7 was narrowed during prosecution by an amendment filed on September 16, 2019 in response to the examiner's § 102 rejection based on Takahashi. The amendment was made expressly to overcome the prior-art rejection — that is, for a reason related to patentability. Under Festo, this triggers the presumption that the patentee surrendered all territory between the original broader claim language and the narrowed "coherent detection module" limitation.

The scope of the surrender territory includes coherent detection architectures that would have been covered by the original broader claim language (which encompassed any coherent detection approach on a silicon substrate) but that do not satisfy the narrowed claim language requiring a "unitary coherent detection module" with the specific structural configuration of optical hybrid, balanced photodetectors, and transimpedance amplifier array integrated on the SOI substrate. Lumicore's distributed coherent detection architecture — which performs coherent detection but distributes the functional components rather than integrating them as a unitary module — arguably falls within this surrender territory.

Lumicore's defense counsel at Hawthorne Vega & Cruz LLP will undoubtedly argue that the Festo presumption is dispositive and that OWS cannot assert DOE with respect to the very limitation that was narrowed to secure allowance of the claims. This is a serious argument that must be confronted directly in our DOE contentions and in any subsequent briefing.

2. Rebuttal Arguments — Festo Exceptions. ##### a. Tangential Relation Exception (Primary Argument)

Our primary argument for overcoming the Festo presumption on the '193 Patent claims is the tangential relation exception. Under this exception, the presumption of surrender is rebutted where "the rationale underlying the amendment [bore] no more than a tangential relation to the equivalent in question." Festo, 535 U.S. at 740–41.

Here, the rationale for the narrowing amendment was to distinguish the claimed coherent detection module from Takahashi's WDM multiplexer/demultiplexer architecture with an integrated coherent receiver. The applicant's September 16, 2019 response argued that Takahashi's coherent receiver was structurally distinct because it was part of a WDM multiplexing/demultiplexing scheme and lacked the "unitary" module configuration recited in the amended claims. The core of the argument for patentability over Takahashi was that the claimed invention provided a self-contained, unitary coherent detection module distinct from Takahashi's distributed WDM receiver architecture.

The equivalent now at issue — Lumicore's distributed coherent detection architecture — is not the same type of design as Takahashi's. Takahashi's reference describes a fundamentally different system: a dense WDM silicon photonic multiplexer/demultiplexer with coherent receiver functionality integrated into a multi-channel WDM architecture. Lumicore's accused products, by contrast, are single-channel (or multi-lane, non-WDM) coherent transceivers that employ a distributed layout of coherent detection components on a silicon photonic die. The reason Lumicore's products do not literally infringe — the distributed (rather than unitary) arrangement of the optical hybrid, photodetectors, and TIA — is a different structural variation than the one that motivated the narrowing amendment, which was focused on distinguishing the claimed invention from Takahashi's multi-channel WDM receiver architecture.

In other words, the amendment was directed at distinguishing the type of system (unitary coherent module vs. WDM receiver architecture) rather than the specific spatial arrangement of coherent detection components that makes Lumicore's design different from the literal claim language. The rationale for the amendment — addressing Takahashi's WDM-based coherent receiver — bore only a tangential relation to the specific equivalent at issue — Lumicore's distributed single-channel coherent detection architecture.

The Federal Circuit's decision in Insituform Technologies, Inc. v. CAT Contracting, Inc., 385 F.3d 1360, 1370 (Fed. Cir. 2004), supports this argument. In Insituform, the court held that the tangential relation exception applied where the amendment was made to address a prior-art reference that was structurally different from the accused equivalent, even though the amendment narrowed the same claim limitation. The court emphasized that the inquiry focuses on "the relationship between the reason for the amendment and the alleged equivalent," not merely on whether the same limitation was amended. Id.

This is our strongest ground for overcoming estoppel, and we recommend that it be the lead argument in the P.R. 3-1 Infringement Contentions, in Dr. Ashford's expert report, and in any opposition to summary judgment.

##### b. Unforeseeability Exception (Secondary Argument)

As a secondary argument, we assess whether Lumicore's specific modified coherent detection architecture was unforeseeable at the time of the narrowing amendment (September 16, 2019). Under the unforeseeability exception, the Festo presumption is rebutted if "the equivalent may have been unforeseeable at the time of the amendment." Festo, 535 U.S. at 740.

The Federal Circuit has interpreted this exception to apply most readily to "later-developed technologies" that were not known or available at the time of prosecution. Festo, 344 F.3d at 1369 (on remand). The exception is more difficult to establish where the equivalent is based on technology that was known in the art at the time of the amendment.

There is a plausible argument that Lumicore's specific distributed coherent detection architecture, with its particular MMI coupler network geometry and flip-chip TIA integration approach, was not foreseeable in September 2019. Lumicore's design documents indicate that the specific distributed architecture was developed internally at Lumicore between 2020 and 2021 (Bates Nos. LUMI-00042187 through LUMI-00042203), and the earliest published description of a comparable architecture in the academic literature appears to date from late 2020. However, the general concept of distributing coherent detection components across a photonic die (rather than integrating them in a compact unitary module) was known in the silicon photonics research community as early as 2017–2018. Several academic papers from this period describe distributed coherent receiver architectures on silicon photonic platforms, although none employ the specific design used by Lumicore.

This distinction — between the general concept of a distributed architecture (known at the time) and the specific design implemented by Lumicore (arguably not foreseeable) — presents a factual question that requires further development through expert analysis. We recommend that Dr. Ashford address this issue in her expert report, with a thorough survey of the state of the art as of September 2019 to determine whether the specific structural variation employed by Lumicore was foreseeable at that time.

Assessment: This argument is viable but weaker than the tangential-relation exception. If the general concept of distributed coherent detection was known in the art at the time of the amendment, Lumicore will argue that the specific variation was foreseeable, even if it had not yet been implemented in a commercial product. We recommend developing this argument fully but presenting it as a secondary basis for overcoming estoppel, not as the primary ground.

##### c. Other Reason Exception

The third Festo exception — that "there [is] some other reason suggesting that the patentee could not reasonably be expected to have described the insubstantial substitute in question," Festo, 535 U.S. at 740 — is narrowly applied by the Federal Circuit and is generally considered a "safety valve" for cases that do not fit neatly into the unforeseeability or tangential-relation categories. See Festo, 344 F.3d at 1369.

We have identified one potential basis for this exception: the argument that the specific structural difference between the claimed unitary module and the accused distributed architecture is a difference of degree (spatial layout) rather than of kind (detection mechanism), and that it would have been unreasonable to expect the applicant to have described every possible spatial layout of the coherent detection components during prosecution. The claim already describes the functional elements (optical hybrid, balanced photodetectors, TIA); the narrowing amendment focused on the "unitary" integration of these elements, and it would have been impracticable for the applicant to catalogue all possible non-unitary spatial distributions as equivalents.

Assessment: This argument has some force but is unlikely to succeed on its own. The Federal Circuit has generally been skeptical of arguments that the applicant could not have drafted broader claim language to capture the equivalent. We recommend presenting this argument only as a supplement to the tangential-relation and unforeseeability exceptions, not as a standalone basis for overcoming estoppel.

3. Overall Estoppel Risk Assessment for '193 Claims 3 and 7. Risk Level: Moderate.

The Festo presumption clearly applies because the "coherent detection module" limitation was narrowed during prosecution to overcome a § 102 rejection. The tangential-relation exception is viable and well-supported: the amendment was directed at distinguishing over Takahashi's WDM-based coherent receiver architecture, which is a structurally different design from Lumicore's distributed single-channel coherent detection architecture. The unforeseeability exception provides secondary support, though its strength depends on further factual development regarding the state of the art as of September 2019.

Defense counsel Kate Rourke at Hawthorne Vega & Cruz LLP will argue that the amendment directly targeted the "coherent detection module" limitation — the very limitation on which OWS now asserts DOE — and that estoppel should therefore foreclose DOE on that limitation. This is a legitimate argument, and the Court could find it persuasive. However, the Festo inquiry does not simply ask whether the same limitation was amended; it asks whether the rationale for the amendment bears on the specific equivalent at issue. The Supreme Court in Festo made clear that "the patentee's decision to narrow his claims through amendment may not be presumed to be an admission that the territory surrendered in an amendment is fair game for all purposes of patent law." 535 U.S. at 737.

We recommend proceeding with DOE contentions on Claims 3 and 7, leading with the tangential-relation argument. Dr. Ashford's expert report should clearly articulate why the amendment was directed at a different aspect of the prior art (Takahashi's WDM architecture) than the equivalent at issue (Lumicore's distributed coherent detection layout). We should also prepare a contingency opposition brief to Lumicore's anticipated summary judgment motion on DOE/prosecution history estoppel, which we expect to be filed in or around September 2025 in advance of the dispositive motions deadline.

IV. Analysis of '015 Patent — Claim 12

A. Claim Language

The relevant claims of the '015 Patent are set forth in full in Appendix D. Claim 12 is a dependent claim. For purposes of this analysis, we reproduce the pertinent claim language below.

Claim 9 (independent, parent claim) recites:

9. A polarization-diverse grating coupler for a silicon photonic transceiver, comprising:

a silicon waveguide layer disposed on a buried oxide layer of a silicon-on-insulator substrate;

a grating structure patterned in the silicon waveguide layer and configured to receive an optical signal from an optical fiber positioned above the grating structure;

a first output waveguide configured to carry a first polarization component of the optical signal; and

a second output waveguide configured to carry a second polarization component of the optical signal, the second polarization component being orthogonal to the first polarization component,

wherein the grating structure is configured to simultaneously couple and separate the first polarization component and the second polarization component from the optical signal into the first output waveguide and the second output waveguide, respectively.

Claim 10 (dependent on Claim 9) recites:

10. The polarization-diverse grating coupler of claim 9, wherein the grating structure comprises a periodic pattern of scattering elements configured to diffract the optical signal into the silicon waveguide layer.

Claim 12 (dependent on Claim 10, and through it on Claim 9) recites:

12. The polarization-diverse grating coupler of claim 10, wherein the grating structure comprises a two-dimensional sub-wavelength grating having a first periodic pattern along a first axis and a second periodic pattern along a second axis perpendicular to the first axis, the first periodic pattern having a first period selected to couple the first polarization component, and the second periodic pattern having a second period selected to couple the second polarization component, wherein the first period and the second period are each less than an operating wavelength of the optical signal divided by an effective refractive index of the silicon waveguide layer.

The "two-dimensional sub-wavelength grating" limitation is the focus of the DOE analysis for Claim 12. The '015 Patent specification describes this structure in detail at column 11, lines 4–48, and column 12, lines 1–22. The specification explains that the two-dimensional sub-wavelength grating "enables simultaneous coupling and polarization splitting in a single grating element by exploiting the anisotropic diffraction properties of a two-dimensional periodic structure with sub-wavelength feature sizes." '015 Patent, col. 11, ll. 8–12. The specification further states that "the sub-wavelength nature of the grating periods ensures that no higher-order diffraction modes propagate in the waveguide layer, thereby achieving high coupling efficiency and low polarization-dependent loss." Id., col. 11, ll. 34–38.

B. Prosecution History of the '015 Patent — The "Two-Dimensional Sub-Wavelength Grating" Amendment

The prosecution history of the '015 Patent is directly relevant to the DOE analysis because the "two-dimensional sub-wavelength grating" limitation was added to the claims by amendment during prosecution. The relevant prosecution history excerpts are included in Appendix B; we summarize the key events below.

On February 4, 2021, the PTO issued a non-final Office Action rejecting the original claims of the '015 Patent under 35 U.S.C. § 103 as obvious over a combination of two references: U.S. Patent No. 8,412,007 (Bergmann, "Polarization-Independent Grating Coupler for Silicon Waveguides," issued April 2, 2013) and International Patent Application PCT/EP2015/071234 (Simonsen, "Integrated Optical Coupler with Polarization Diversity," filed September 17, 2015). The examiner found that Bergmann disclosed a grating coupler with a periodic scattering pattern for coupling light from a fiber into a silicon waveguide, and that Simonsen disclosed a polarization-diverse coupling approach with separate output waveguides for orthogonal polarization states. The examiner concluded that it would have been obvious to combine the teachings of Bergmann and Simonsen to arrive at the claimed polarization-diverse grating coupler.

Importantly, Bergmann's grating coupler employed a one-dimensional grating structure — a periodic pattern along a single axis — that did not inherently provide polarization splitting. Simonsen's coupler achieved polarization diversity through a different geometry — a tapered waveguide structure downstream of a conventional grating coupler, rather than through a two-dimensional grating with inherent polarization-splitting properties.

On August 12, 2021, the applicant (through Dr. Martin Chu at Cascade IP Advisors) filed a response to the Office Action that included an amendment to the dependent claim that became Claim 12. The amendment added the "two-dimensional sub-wavelength grating" modifier to the grating limitation. The original claim language covered grating couplers with a periodic pattern of scattering elements generally, without specifying the two-dimensional geometry or the sub-wavelength dimensional constraint. The amendment introduced the requirement of (1) a two-dimensional grating with periodic patterns along two perpendicular axes, (2) with each period selected to couple a respective polarization component, and (3) with each period being less than the operating wavelength divided by the effective refractive index (i.e., sub-wavelength).

In the accompanying remarks, Dr. Chu argued that the amended claim language distinguished over Bergmann and Simonsen because:

"Neither Bergmann nor Simonsen, alone or in combination, teaches or suggests a two-dimensional sub-wavelength grating with the specific structural and functional characteristics recited in amended Claim 12. Bergmann discloses only a one-dimensional grating coupler without polarization-splitting capability, and Simonsen achieves polarization diversity through a tapered waveguide structure downstream of the grating, not through the grating itself. The two-dimensional sub-wavelength grating of amended Claim 12 achieves simultaneous coupling and polarization splitting in a single grating element by virtue of its two-dimensional periodic structure with sub-wavelength feature sizes — a feature absent from both references."

The examiner allowed the claims. A Notice of Allowance was issued on March 22, 2022, and the patent issued on August 2, 2022.

The narrowing amendment — adding the "two-dimensional sub-wavelength" modifier — was made to overcome the § 103 obviousness rejection based on Bergmann and Simonsen. This was an amendment for reasons of patentability, triggering the Festo presumption of prosecution history estoppel. The scope of surrender encompasses the territory between the original broader grating coupler limitation (which covered any grating coupler with a periodic pattern of scattering elements) and the narrowed limitation requiring a "two-dimensional sub-wavelength grating" with the specific structural characteristics described in Claim 12.

C. Lumicore's Accused Design — How It Differs from the Literal Claim

The Benchmark teardown reports for the LumiLink 800G-DR8 (June 2023) and the LumiEdge 1600G-LR (February 2024) reveal that Lumicore's accused products employ a grating coupler structure that achieves polarization-diverse coupling of light from a fiber into the silicon photonic waveguide layer. However, the accused grating structure may not literally satisfy the "two-dimensional sub-wavelength grating" limitation of Claim 12.

Specifically, Benchmark's analysis — based on scanning electron microscopy (SEM) imaging and focused ion beam (FIB) cross-sectioning of the accused grating structures — indicates that Lumicore's grating coupler employs a quasi-periodic two-dimensional grating geometry in which the grating periods along one of the two axes slightly exceed the sub-wavelength threshold defined in the claim (i.e., the period along one axis is slightly greater than the operating wavelength divided by the effective refractive index of the silicon waveguide layer). The grating is two-dimensional in the sense that it has periodic structures along two perpendicular axes, but the dimensional constraint of "sub-wavelength" (as defined in the claim) may not be strictly satisfied along one axis.

Kevin Ng's Benchmark report for the LumiLink 800G-DR8 states:

"SEM imaging of the grating coupler die region reveals a two-dimensional grating pattern with nominal periods of approximately 580 nm along the x-axis and approximately 640 nm along the y-axis. Given the operating wavelength of 1310 nm and the estimated effective refractive index of the SOI waveguide layer of approximately 2.1, the sub-wavelength threshold is approximately 624 nm (1310 nm / 2.1). The x-axis period of 580 nm is below this threshold, but the y-axis period of 640 nm slightly exceeds it."

Benchmark Teardown Report, LumiLink 800G-DR8, at 51. The LumiEdge 1600G-LR exhibits a similar grating structure with comparable dimensional parameters. See Benchmark Teardown Report, LumiEdge 1600G-LR, at 33.

The fact that the y-axis period (approximately 640 nm) slightly exceeds the sub-wavelength threshold (approximately 624 nm) means that Lumicore's accused grating may not literally satisfy the claim requirement that "the first period and the second period are each less than an operating wavelength of the optical signal divided by an effective refractive index of the silicon waveguide layer." This is the specific structural difference that motivates the DOE analysis for Claim 12.

Despite this dimensional departure, the accused grating structure accomplishes the same polarization-diverse coupling function as the claimed two-dimensional sub-wavelength grating. The grating achieves simultaneous coupling and polarization splitting of both TE and TM modes from the fiber into separate output waveguides. The slight exceedance of the sub-wavelength threshold along one axis does not prevent the grating from operating as a polarization-diverse coupler; it merely means that a low-power higher-order diffraction mode may propagate along that axis, which Lumicore suppresses through additional design features (apodization and absorptive boundary structures at the waveguide edges, per Lumicore design files at Bates No. LUMI-00078334).

D. Function-Way-Result Analysis

We apply the function-way-result test to the "two-dimensional sub-wavelength grating" limitation of Claim 12 as applied to Lumicore's accused products.

1. Function. The claimed "two-dimensional sub-wavelength grating" functions as a polarization-splitting coupling element that simultaneously couples both TE and TM polarization modes from an optical fiber into the silicon photonic integrated circuit, directing the two orthogonal polarization components into separate output waveguides. This dual-function coupling-and-splitting operation is achieved in a single grating element, eliminating the need for a separate polarization beam splitter downstream of the grating coupler.

Lumicore's accused grating coupler performs substantially the same function. The accused two-dimensional grating structure simultaneously couples both TE and TM polarization modes from the input fiber into the

PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT
MEMORANDUM
PRIVILEGED AND CONFIDENTIAL — ATTORNEY-CLIENT PRIVILEGE / ATTORNEY WORK PRODUCT DOCTRINE
This memorandum was prepared by attorneys at Calderwood & Harkness LLP at the direction of, and for the benefit of, our client Optiwave Semiconductor Corp. ("OWS" or "Client") in connection with pending litigation and is protected by the attorney-client privilege and the attorney work product doctrine. This memorandum contains mental impressions, conclusions, opinions, and legal theories of the attorneys involved in this litigation. It is not intended for disclosure to any third party and should not be disclosed without the express written consent of the undersigned.
Right-click to update Table of Contents
I. Executive Summary
This memorandum analyzes the viability of asserting infringement under the doctrine of equivalents ("DOE") for three specific patent claims asserted by Optiwave Semiconductor Corp. ("OWS") against Lumicore Technologies, Inc. ("Lumicore") in the above-captioned matter: Claims 3 and 7 of U.S. Patent No. 10,891,193 (the "'193 Patent") and Claim 12 of U.S. Patent No. 11,402,015 (the "'015 Patent"). This analysis is undertaken in advance of OWS's obligation to serve its P.R. 3-1 Infringement Contentions on September 16, 2024, pursuant to the Court's scheduling order, and is intended to inform the litigation team's decision-making regarding which DOE theories to include in those contentions.
OWS accuses three Lumicore product families of infringing the asserted claims: the LumiLink 400G-ZR+ (accused of infringing all three claims), the LumiLink 800G-DR8 (accused of infringing all three claims), and the LumiEdge 1600G-LR (accused of infringing '193 Patent Claims 3 and 7 and '015 Patent Claim 12). These products represent a combined FY 2023 revenue of approximately $264.9 million and are central to Lumicore's silicon photonics transceiver product line.
The DOE analysis in this memorandum focuses on two specific claim limitations where literal infringement may be contested: (1) the "coherent detection module" limitation in '193 Patent Claims 3 and 7, and (2) the "two-dimensional sub-wavelength grating" limitation in '015 Patent Claim 12. In both instances, Lumicore's accused products employ structures that depart from the literal claim language as narrowed during prosecution but that perform substantially the same function, in substantially the same way, to achieve substantially the same result as the claimed elements.
Bottom-Line Assessment. DOE contentions are supportable for all three claims. However, prosecution history estoppel presents a moderate risk with respect to '193 Patent Claims 3 and 7 and a lower-but-non-trivial risk with respect to '015 Patent Claim 12. In both cases, the claim limitations at issue were narrowed during prosecution to overcome prior-art rejections, triggering the Festo presumption of surrender. We believe this presumption is rebuttable under the "tangential relation" exception recognized by the Supreme Court in Festo Corp. v. Fenmore & Co., 535 U.S. 722 (2002), and we recommend proceeding with that exception as the primary argument for overcoming estoppel on both patents. The unforeseeability exception provides a secondary basis, particularly for '015 Patent Claim 12.
We recommend proceeding with DOE contentions on all three claims in the P.R. 3-1 Infringement Contentions. This belt-and-suspenders approach preserves the maximum scope of OWS's infringement theories, provides strategic flexibility through Markman and dispositive motions, and ensures that OWS's damages case captures the full scope of Lumicore's infringing activity. The detailed recommendations are set forth in Section VI of this memorandum.
II. Background
A. The Asserted Patents and Claims
1. The '193 Patent — U.S. Patent No. 10,891,193. The '193 Patent, titled "High-Bandwidth Coherent Optical Transceiver Integrated on Silicon Substrate," was filed on April 7, 2017 and issued on January 12, 2021. The original assignee of record is Optiwave Semiconductor Corp. The '193 Patent claims the benefit of U.S. Provisional Application No. 62/499,117, filed April 7, 2017, which is a common provisional application shared with the '015 Patent discussed below. The '193 Patent was prosecuted by Dr. Martin Chu of Cascade IP Advisors, LLC (1550 Crystal Drive, Suite 600, Arlington, VA 22202), who served as OWS's primary patent prosecution counsel during the relevant period. The sole named inventor is Dr. Jian-Wei Huang, OWS's Chief Technology Officer, who conceived of the invention during OWS's development of its next-generation silicon photonic transceiver platform.
The '193 Patent is directed generally to an integrated coherent optical transceiver fabricated on a silicon substrate, enabling high-bandwidth data transmission and reception using coherent detection techniques. The patent specification describes, in considerable detail, the architecture of the transceiver, including the integration of a transmitter path and a receiver path on a common silicon-on-insulator (SOI) substrate. A central feature of the receiver path is the "coherent detection module," which the specification describes as a key structural component responsible for mixing received optical signals with a local oscillator (LO) to extract both phase and amplitude information from the incoming signal.
The claims at issue for DOE analysis in this memorandum are Claim 3 (which depends directly from independent Claim 1) and Claim 7 (which depends from Claim 1 through intervening Claim 5). Both Claims 3 and 7 recite the "coherent detection module" limitation, which is the focus of the DOE analysis. As discussed in detail in Section III below, Lumicore's accused products may not literally satisfy this limitation because they employ a modified coherent detection architecture that departs from the specific structural and functional features recited in the amended claim language. However, the accused architecture performs the same coherent detection function by the same fundamental operating principle, achieving the same result, making it a strong candidate for DOE analysis.
2. The '015 Patent — U.S. Patent No. 11,402,015. The '015 Patent, titled "Polarization-Diverse Grating Coupler for Silicon Photonic Transceivers," was filed on August 30, 2019 and issued on August 2, 2022. Like the '193 Patent, the '015 Patent was originally assigned to Optiwave Semiconductor Corp. and claims the benefit of the same U.S. Provisional Application No. 62/499,117, filed April 7, 2017. The '015 Patent was also prosecuted by Dr. Martin Chu at Cascade IP Advisors, LLC.
The '015 Patent is directed to a polarization-diverse grating coupler designed for use in silicon photonic transceivers. The specification describes the challenges of coupling light from a standard optical fiber into a silicon photonic integrated circuit (PIC) while maintaining sensitivity to both transverse-electric (TE) and transverse-magnetic (TM) polarization modes. The patent teaches the use of a two-dimensional sub-wavelength grating structure that simultaneously couples both polarization states from the fiber into the waveguide layer, splitting them into separate waveguide paths for independent processing. This approach is described as superior to prior-art solutions that required separate grating couplers or polarization-diversity schemes external to the coupling element.
The claim at issue for DOE analysis is Claim 12, which is a dependent claim. The key limitation subject to DOE analysis is the "two-dimensional sub-wavelength grating" limitation. As discussed in Section IV below, Lumicore's accused products employ a grating structure that achieves the same polarization-diverse coupling function but may depart from the literal "two-dimensional sub-wavelength" geometry specified in the amended claim language. The accused grating structure achieves its polarization-diversity function through a structurally analogous but not literally identical grating geometry, making it a viable candidate for DOE.
B. Accused Products
OWS accuses three Lumicore product families of infringing the asserted claims. Each of these products is a pluggable optical transceiver module designed for high-speed data center interconnect applications and is manufactured using Lumicore's proprietary silicon photonic platform. The relevant products and their commercial significance are as follows:
LumiLink 400G-ZR+. Lumicore launched the LumiLink 400G-ZR+ in the third quarter of 2022 as a coherent pluggable transceiver module designed for long-haul and metro data center interconnect applications. The 400G-ZR+ implements a coherent transceiver architecture on a silicon photonic die, incorporating both a coherent transmitter and a coherent receiver in a single QSFP-DD form factor. Lumicore reported FY 2023 revenue of approximately $94.8 million for the 400G-ZR+ product line. The LumiLink 400G-ZR+ is accused of infringing all three claims at issue ('193 Patent Claims 3 and 7, and '015 Patent Claim 12).
LumiLink 800G-DR8. Launched in the first quarter of 2023, the LumiLink 800G-DR8 is a short-reach (500m) transceiver module designed for intra-data-center applications at 800 Gbps aggregate bandwidth. The 800G-DR8 employs eight lanes of 100G PAM4 signaling on a silicon photonic platform, with a polarization-diverse grating coupler architecture for fiber coupling. Lumicore reported FY 2023 revenue of approximately $128.4 million for the 800G-DR8 product line, making it Lumicore's highest-revenue accused product. The LumiLink 800G-DR8 is accused of infringing all three claims at issue.
LumiEdge 1600G-LR. Lumicore launched the LumiEdge 1600G-LR in the fourth quarter of 2023 as its next-generation long-reach transceiver module, targeting 1.6 Tbps data rates for next-generation data center interconnect. The 1600G-LR is the newest of the accused products and represents Lumicore's entry into the emerging 1.6T market segment. Lumicore reported FY 2023 revenue of approximately $41.7 million for the 1600G-LR product line (reflecting only one quarter of commercial availability). The LumiEdge 1600G-LR is accused of infringing '193 Patent Claims 3 and 7 and '015 Patent Claim 12.
The total accused product revenue for FY 2023 is $264.9 million across all three product families. This figure is expected to increase substantially in FY 2024 as the 800G-DR8 and 1600G-LR products gain broader market adoption.
Benchmark Semiconductor Analysis, LLC ("Benchmark") was retained to perform physical teardowns and structural analysis of all three accused products. Kevin Ng, Lead Engineer at Benchmark, supervised all teardown activities. The initial teardown report for the LumiLink 400G-ZR+ was delivered on December 12, 2022. Follow-on teardown reports were completed for the LumiLink 800G-DR8 (delivered June 2023) and the LumiEdge 1600G-LR (delivered February 2024). The Benchmark teardown data, combined with Lumicore design documents produced in discovery, form the factual foundation for the DOE analysis in this memorandum. Benchmark's findings on the relevant structural features are summarized in Appendix C.
C. Legal Standard for the Doctrine of Equivalents
The doctrine of equivalents is a judicially created doctrine that allows a patentee to claim infringement even when the accused product does not literally satisfy every limitation of a patent claim. The Supreme Court reaffirmed the continuing vitality of the doctrine in Warner-Jenkinson Co. v. Hilton Davis Chemical Co., 520 U.S. 17 (1997), holding that the doctrine of equivalents remains an important part of patent law and is applied on an element-by-element basis.
Under the All-Elements Rule, the doctrine of equivalents must be applied to individual claim limitations, not to the claim as a whole. Warner-Jenkinson, 520 U.S. at 29. A patentee must demonstrate that each element of the claim is met in the accused product, either literally or by an equivalent. The doctrine cannot be applied in a manner that effectively vitiates (reads out entirely) a claim limitation. Id. at 39 n.8.
Courts have recognized two formulations for determining whether an element of an accused product is equivalent to a claim limitation:
1. The Function-Way-Result Test (derived from Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 608 (1950)): Under this test, an accused element is equivalent to a claim limitation if it performs substantially the same function, in substantially the same way, to achieve substantially the same result as the claimed element.
2. The Insubstantial Differences Test: Under this alternative formulation, infringement under the doctrine of equivalents is found when the differences between the claim limitation and the accused element are insubstantial. See Warner-Jenkinson, 520 U.S. at 40.
This memorandum primarily applies the function-way-result test, as is standard in Eastern District of Texas practice and consistent with Judge Gilstrap's prior rulings in patent cases. See, e.g., Omega Patents, LLC v. CalAmp Corp., No. 2:13-cv-00726-JRG, 2017 WL 3425704 (E.D. Tex. Aug. 9, 2017) (Gilstrap, J.) (applying function-way-result test in evaluating DOE contentions at summary judgment). The insubstantial-differences test may provide additional support but is not the primary framework for analysis.
Critically, equivalence is assessed at the time of infringement, not at the time the patent was issued. Warner-Jenkinson, 520 U.S. at 37. This temporal requirement means that technologies developed after the patent's issuance can still be captured by the doctrine of equivalents if they are equivalent to the claimed element at the time of the alleged infringement. This principle may be relevant here, particularly with respect to Lumicore's design techniques that were implemented in products launched in 2022 and 2023.
D. Legal Standard for Prosecution History Estoppel
The doctrine of prosecution history estoppel operates as a limitation on the doctrine of equivalents. Where a patentee has narrowed the scope of a claim during prosecution for reasons related to patentability, the patentee may be estopped from later recapturing through the doctrine of equivalents the subject matter that was surrendered by the narrowing amendment.
The Supreme Court set forth the governing framework for prosecution history estoppel in Festo Corp. v. Fenmore & Co., 535 U.S. 722 (2002). Under Festo, when a claim limitation is narrowed by amendment during prosecution for a reason related to patentability — such as to overcome a prior-art rejection under 35 U.S.C. § 102 or § 103 — a presumption arises that the patentee has surrendered all territory between the original claim limitation and the amended claim limitation. Festo, 535 U.S. at 740. This presumption applies regardless of whether the amendment was made at the suggestion of the examiner or was voluntarily offered by the applicant; what matters is whether the amendment was made for a reason related to patentability. Id. at 736–37.
The Festo presumption of surrender is, however, rebuttable. The Supreme Court recognized three circumstances under which a patentee may overcome the presumption:
1. The Unforeseeability Exception: The equivalent at issue was unforeseeable at the time of the narrowing amendment. Festo, 535 U.S. at 740–41. If the alleged equivalent was based on technology that did not exist or was not known in the relevant art at the time of the amendment, the patentee could not reasonably have been expected to have claimed it, and the presumption of surrender does not apply.
2. The Tangential Relation Exception: The rationale underlying the narrowing amendment bore only a tangential relation to the equivalent in question. Id. at 740–41. If the reason the applicant narrowed the claim had nothing (or only a peripheral connection) to do with the specific equivalent now at issue, the patentee should not be foreclosed from asserting equivalence with respect to that particular variation.
3. The "Other Reason" Exception: There is some other reason suggesting that the patentee could not reasonably have been expected to have described the insubstantial substitute in question. Id. This exception serves as a catch-all but is narrowly applied by the Federal Circuit. See Festo Corp. v. Fenmore & Co., 344 F.3d 1359, 1369 (Fed. Cir. 2003) (en banc) (on remand).
The burden of overcoming the Festo presumption rests squarely on the patentee. Festo, 535 U.S. at 740. This is a meaningful burden, and the patentee must present specific, persuasive evidence and argument to rebut the presumption. The Federal Circuit has emphasized that "[t]he patentee bears the burden of showing that the amendment does not surrender the particular equivalent in question." Honeywell Int'l, Inc. v. Hamilton Sundstrand Corp., 370 F.3d 1131, 1142 (Fed. Cir. 2004) (en banc).
It is also important to note that the Festo framework applies to voluntary amendments, not just those made in direct response to examiner rejections. Festo, 535 U.S. at 736–37. The inquiry is whether the amendment was made "for a substantial reason related to patentability," which includes amendments offered proactively by the applicant to advance prosecution, so long as they narrow the claim scope and relate to patentability considerations.
III. Analysis of '193 Patent — Claims 3 and 7
A. Claim Language
The relevant claims of the '193 Patent are set forth in full in Appendix D. For purposes of this analysis, we reproduce the pertinent claim language below.
Claim 1 (independent) recites:
1. An integrated coherent optical transceiver, comprising:
a silicon-on-insulator substrate;
a transmitter path integrated on the silicon-on-insulator substrate, the transmitter path comprising a laser source optically coupled to a Mach-Zehnder modulator array configured to modulate an optical carrier signal with a data signal;
a receiver path integrated on the silicon-on-insulator substrate, the receiver path comprising:
an input waveguide configured to receive a modulated optical signal from an optical fiber;
a local oscillator source integrated on the silicon-on-insulator substrate; and
a signal processing unit configured to recover a data signal from the modulated optical signal;
wherein the transmitter path and the receiver path are monolithically integrated on the silicon-on-insulator substrate and share at least one common optical component.
Claim 3 (dependent on Claim 1) recites:
3. The integrated coherent optical transceiver of claim 1, wherein the receiver path further comprises a coherent detection module disposed between the input waveguide and the signal processing unit, the coherent detection module comprising:
an optical hybrid configured to mix the modulated optical signal with a local oscillator signal from the local oscillator source to produce at least four mixed optical signals;
a plurality of balanced photodetectors, each balanced photodetector configured to receive a respective pair of the mixed optical signals and to produce a differential electrical signal proportional to an optical phase difference; and
a transimpedance amplifier array electrically coupled to the plurality of balanced photodetectors,
wherein the optical hybrid, the plurality of balanced photodetectors, and the transimpedance amplifier array are integrated on the silicon-on-insulator substrate as a unitary coherent detection module.
Claim 7 (dependent on Claim 1 through Claim 5) recites:
7. The integrated coherent optical transceiver of claim 5, wherein the coherent detection module is configured to detect both in-phase (I) and quadrature (Q) components of the modulated optical signal by producing at least eight mixed optical signals from the optical hybrid, and wherein the coherent detection module further comprises a polarization beam splitter integrated upstream of the optical hybrid to separate orthogonal polarization states of the modulated optical signal into two polarization paths, each polarization path directed to a respective optical hybrid.
The "coherent detection module" limitation, which is the focus of this DOE analysis, appears in both Claims 3 and 7. The specification of the '193 Patent describes the coherent detection module in detail at column 7, lines 12–55, and column 8, lines 1–34. The specification states that "[t]he coherent detection module 300 is a key component of the receiver path, enabling the extraction of both phase and amplitude information from the incoming optical signal through coherent mixing with a local oscillator." '193 Patent, col. 7, ll. 12–16. The specification further describes the specific structural arrangement of the optical hybrid, balanced photodetectors, and transimpedance amplifier array as a "unitary" module integrated on the SOI substrate. Id., col. 7, ll. 23–42.
B. Prosecution History of the '193 Patent — The "coherent Detection Module" Amendment
The prosecution history of the '193 Patent is directly relevant to the DOE analysis because the "coherent detection module" limitation was narrowed during prosecution in a manner that triggers the Festo presumption of prosecution history estoppel. The relevant prosecution history excerpts are included in Appendix A; we summarize the key events below.
On March 18, 2019, the U.S. Patent and Trademark Office ("PTO") issued a non-final Office Action rejecting original Claim 1 and all dependent claims (including the predecessors to current Claims 3 and 7) under 35 U.S.C. § 102(a)(1) based on Takahashi et al., "Dense WDM Silicon Photonic MUX with Integrated Coherent Receiver," IEEE Photonics Technology Letters, Vol. 26, No. 6, pp. 600–603 (March 15, 2014) ("Takahashi"). The examiner found that Takahashi disclosed an integrated silicon photonic transceiver with a coherent receiver module that anticipated the originally filed claims.
In the original claim language, the "coherent detection module" limitation was drafted more broadly. The original claim language did not include the requirement that the optical hybrid, balanced photodetectors, and transimpedance amplifier array be integrated on the SOI substrate "as a unitary coherent detection module," nor did it specify the structural arrangement of the optical hybrid to produce "at least four mixed optical signals." The original language encompassed any coherent detection approach integrated on a silicon substrate, without the structural and functional specificity that appears in the issued claims.
On September 16, 2019, the applicant (through Dr. Martin Chu at Cascade IP Advisors) filed a response to the Office Action that included amendments to the pending claims. Specifically, the response amended the "coherent detection module" limitation to add the structural and functional requirements present in the issued claim language — i.e., requiring the specific configuration of the optical hybrid producing at least four mixed optical signals, the balanced photodetectors receiving respective pairs of mixed optical signals, the transimpedance amplifier array, and the integration of these components as a "unitary coherent detection module" on the SOI substrate. In the accompanying remarks, Dr. Chu argued that the amended claim language distinguished over Takahashi because Takahashi's coherent receiver lacked the specific structural integration and unitary module configuration recited in the amended claims. The remarks stated, in relevant part:
"Amended Claim 1 and its dependent claims are patentably distinct from Takahashi. Takahashi describes a coherent receiver section within a WDM multiplexer/demultiplexer structure, but does not teach or suggest the specific unitary coherent detection module configuration recited in the amended claims, in which the optical hybrid, balanced photodetectors, and transimpedance amplifier array are integrated on the SOI substrate as a single, self-contained detection module. Takahashi's receiver architecture distributes these functions across separate, non-unitary components that are not co-integrated in the manner claimed."
The examiner subsequently allowed the claims. A Notice of Allowance was issued on July 8, 2020, and the patent issued on January 12, 2021.
The critical point for DOE analysis is that the narrowing amendment to the "coherent detection module" limitation was made expressly to overcome the § 102 rejection based on Takahashi — that is, for a reason related to patentability. This triggers the Festo presumption that the patentee surrendered the territory between the original broader claim language and the narrowed "coherent detection module" limitation as amended. The scope of surrender encompasses coherent detection approaches that would have been covered by the original broader claim language but that fall outside the narrowed limitation specifying the unitary module configuration with the particular structural requirements.
C. Lumicore's Accused Design — How It Differs from the Literal Claim
Based on the Benchmark teardown data and Lumicore's design documents produced in discovery (Bates Nos. LUMI-00042187 through LUMI-00042643 and LUMI-00078219 through LUMI-00078456), the accused products — the LumiLink 400G-ZR+, LumiLink 800G-DR8, and LumiEdge 1600G-LR — each employ a coherent detection architecture on a silicon photonic platform. The coherent detection functionality in the accused products accomplishes the same overarching purpose as the "coherent detection module" recited in Claims 3 and 7: extracting phase and amplitude information from an incoming optical signal through coherent mixing with a local oscillator.
However, the accused products' coherent detection architecture differs from the literal claim language in a critical structural respect. Specifically, rather than employing a "unitary coherent detection module" in which the optical hybrid, balanced photodetectors, and transimpedance amplifier array are integrated as a single, self-contained module on the SOI substrate (as required by the amended claim language), Lumicore's accused products employ a distributed coherent detection architecture. In this architecture, the optical hybrid function is performed by a multimode interference (MMI) coupler network that is physically separated from the photodetector array on the die, with the two functional blocks interconnected through a set of intermediate silicon waveguides. The transimpedance amplifier (TIA) array in the accused products is implemented on a separate electronic die that is flip-chip bonded onto the silicon photonic die, rather than being monolithically integrated on the SOI substrate as part of a unitary module.
Kevin Ng's Benchmark teardown report for the LumiLink 400G-ZR+ (December 12, 2022) specifically noted that "[t]he coherent receiver section of the LumiLink 400G-ZR+ employs a distributed architecture in which the optical mixing and photodetection functions are spatially separated on the PIC, with the TIA implemented on a co-packaged CMOS die." Benchmark Teardown Report at 37. Similar observations were made in the follow-on teardown reports for the LumiLink 800G-DR8 (June 2023 report at 42) and the LumiEdge 1600G-LR (February 2024 report at 29).
The specific structural difference — the distributed (rather than unitary) module architecture and the separate TIA die — means that Lumicore's accused products may not literally satisfy the "coherent detection module" limitation as narrowed during prosecution. Lumicore's defense counsel will undoubtedly argue that the accused products lack a "unitary coherent detection module" as required by the claim and, therefore, do not literally infringe. This is the precise gap that the doctrine of equivalents is designed to bridge.
D. Function-Way-Result Analysis
We apply the tripartite function-way-result test to the "coherent detection module" limitation of Claims 3 and 7 as applied to Lumicore's accused products.
1. Function. The claimed "coherent detection module" functions to detect modulated optical signals using coherent detection techniques — specifically, by mixing the incoming modulated optical signal with a local oscillator signal to extract both phase and amplitude information from the received signal. This function is fundamental to coherent optical communication and enables the recovery of complex modulation formats (such as QPSK, 16-QAM, and higher-order formats) that carry information in both the amplitude and phase of the optical carrier.
Lumicore's accused coherent detection architecture performs substantially the same function. The distributed MMI coupler network in the accused products mixes the incoming modulated optical signal with a local oscillator signal to produce mixed optical signals carrying phase information. The balanced photodetectors in the accused products detect these mixed signals and produce differential electrical signals proportional to optical phase differences. The co-packaged TIA array amplifies the resulting electrical signals. In all material respects, the function of the accused architecture is identical to the function of the claimed coherent detection module: coherent detection of optical signals to extract phase and amplitude information.
There is no meaningful dispute regarding function. Both the claimed element and the accused element are coherent detection systems that extract phase and amplitude information from incoming modulated optical signals through coherent mixing with a local oscillator. The accused products' promotional materials, including Lumicore's own product datasheets (Bates No. LUMI-00015832), describe the products as implementing "coherent detection" on a "silicon photonic platform" — language that directly aligns with the function of the claimed limitation.
Conclusion: The accused element performs substantially the same function as the claimed "coherent detection module."
2. Way. The more nuanced question is whether the accused design operates in "substantially the same way" as the claimed coherent detection module. The accused products employ a distributed architecture rather than a unitary module, and the TIA is implemented on a separate electronic die rather than being monolithically integrated on the SOI substrate. Lumicore will argue that these structural differences represent a fundamentally different "way" of achieving coherent detection.
However, the operational principle of the accused architecture is the same as the claimed element. Both the claimed and accused systems rely on the same fundamental mechanism: coherent mixing of the incoming signal light with local oscillator light using an optical hybrid structure (whether implemented as a 90° hybrid or an MMI coupler network), followed by detection of the resulting interference signals using balanced photodetectors, and amplification by transimpedance amplifiers. The physics of the coherent detection process — the interference of signal and LO fields on the photodetector surfaces, the balanced detection to suppress common-mode noise, and the transimpedance amplification of the photocurrent — is identical in both the claimed and accused designs.
The structural difference between a "unitary" module and a "distributed" architecture is, in our analysis, a difference of physical layout and packaging, not a difference in the fundamental mechanism of operation. The MMI coupler network in the accused products performs the same optical mixing function as the optical hybrid in the claimed limitation, using the same physical principle of multimode interference to generate the required phase-shifted optical outputs. The balanced photodetectors in the accused products are germanium-on-silicon (Ge-on-Si) photodiodes — the same photodetector technology described in the '193 Patent specification. The only structural difference is that these components are distributed across a larger area of the PIC and interconnected by waveguides rather than being co-located in a single, compact "unitary" module.
With respect to the TIA placement, while the accused products implement the TIA on a separate CMOS die that is flip-chip bonded to the PIC, the TIA still functions as the final amplification stage of the coherent detection chain. The use of co-packaged or flip-chip-bonded electronic dies for TIA implementation is a design choice driven by the practical difficulty of fabricating high-performance TIAs in a silicon photonic process, and does not alter the fundamental "way" in which coherent detection is achieved.
Benchmark's teardown data supports this analysis. Kevin Ng's reports note that the optical mixing, photodetection, and amplification functions in the accused products operate on the same physical principles as described in the '193 Patent, with the structural differences limited to the spatial distribution and packaging of the functional elements. See Benchmark 400G-ZR+ Teardown Report at 38–39; Benchmark 800G-DR8 Teardown Report at 43–44.
We anticipate that Lumicore will argue that the distributed architecture represents a different "way" because the spatial separation of the MMI coupler network and the photodetector array introduces different optical path lengths, potentially different phase-matching conditions, and different noise characteristics compared to a unitary module. This counterargument has some surface appeal but is ultimately unpersuasive. The differences in optical path length and phase matching are compensated for by Lumicore's waveguide routing design (as documented in Lumicore's own design files, Bates No. LUMI-00042301 at 12), and the noise characteristics of the balanced detection scheme remain fundamentally the same. The accused design does not employ a qualitatively different detection mechanism; it simply distributes the same functional elements across a different physical layout.
Conclusion: The accused element operates in substantially the same way as the claimed "coherent detection module." The differences in physical layout and TIA packaging are differences of degree, not of kind, and do not alter the fundamental operational mechanism of coherent optical detection.
3. Result. Both the claimed "coherent detection module" and the accused coherent detection architecture achieve substantially the same result: high-bandwidth coherent detection of modulated optical signals, with recovery of both in-phase and quadrature components of the signal, enabling the demodulation of complex modulation formats on a silicon-integrated photonic platform.
The performance specifications of the accused products confirm that they achieve the same end result as the claimed invention. The LumiLink 400G-ZR+ achieves 400 Gbps coherent transmission and reception; the LumiLink 800G-DR8 achieves 800 Gbps aggregate bandwidth using coherent-like detection in combination with PAM4 signaling; and the LumiEdge 1600G-LR achieves 1.6 Tbps aggregate bandwidth. In each case, the coherent detection architecture in the accused products recovers phase and amplitude information from the received optical signal, enabling high-spectral-efficiency communication — precisely the result described in the '193 Patent specification for the claimed coherent detection module.
Lumicore's product datasheets (Bates No. LUMI-00015832, LUMI-00015847, LUMI-00015863) describe the products as achieving "coherent detection with full phase recovery" on a "silicon photonic platform," language that directly mirrors the result described in the '193 Patent.
Conclusion: The accused element achieves substantially the same result as the claimed "coherent detection module."
4. Overall DOE Assessment for '193 Claims 3 and 7. Based on the foregoing function-way-result analysis, we conclude that the accused products — the LumiLink 400G-ZR+, LumiLink 800G-DR8, and LumiEdge 1600G-LR — infringe Claims 3 and 7 of the '193 Patent under the doctrine of equivalents with respect to the "coherent detection module" limitation. The accused distributed coherent detection architecture performs substantially the same function (coherent detection to extract phase and amplitude information), in substantially the same way (coherent mixing of signal and LO light using an optical hybrid/MMI coupler, balanced photodetection, and transimpedance amplification), to achieve substantially the same result (high-bandwidth coherent optical signal recovery on a silicon photonic platform) as the claimed unitary coherent detection module.
Dr. Caroline Ashford of the Northfield Institute of Technology, OWS's retained technical expert, is expected to support this analysis in her infringement expert report, which is due on June 15, 2025 under the Court's scheduling order. Dr. Ashford has reviewed the Benchmark teardown data and preliminary Lumicore design documents and has indicated in preliminary discussions that the function-way-result analysis is well-supported by the technical evidence. Her expert report should present this analysis in detail, with reference to specific teardown findings, Lumicore design files, and published technical literature on coherent detection architectures.
E. Prosecution History Estoppel Risk for '193 Claims 3 and 7
1. The Festo Presumption Applies. As discussed in Section III.B above, the "coherent detection module" limitation in Claims 3 and 7 was narrowed during prosecution by an amendment filed on September 16, 2019 in response to the examiner's § 102 rejection based on Takahashi. The amendment was made expressly to overcome the prior-art rejection — that is, for a reason related to patentability. Under Festo, this triggers the presumption that the patentee surrendered all territory between the original broader claim language and the narrowed "coherent detection module" limitation.
The scope of the surrender territory includes coherent detection architectures that would have been covered by the original broader claim language (which encompassed any coherent detection approach on a silicon substrate) but that do not satisfy the narrowed claim language requiring a "unitary coherent detection module" with the specific structural configuration of optical hybrid, balanced photodetectors, and transimpedance amplifier array integrated on the SOI substrate. Lumicore's distributed coherent detection architecture — which performs coherent detection but distributes the functional components rather than integrating them as a unitary module — arguably falls within this surrender territory.
Lumicore's defense counsel at Hawthorne Vega & Cruz LLP will undoubtedly argue that the Festo presumption is dispositive and that OWS cannot assert DOE with respect to the very limitation that was narrowed to secure allowance of the claims. This is a serious argument that must be confronted directly in our DOE contentions and in any subsequent briefing.
2. Rebuttal Arguments — Festo Exceptions. ##### a. Tangential Relation Exception (Primary Argument)
Our primary argument for overcoming the Festo presumption on the '193 Patent claims is the tangential relation exception. Under this exception, the presumption of surrender is rebutted where "the rationale underlying the amendment [bore] no more than a tangential relation to the equivalent in question." Festo, 535 U.S. at 740–41.
Here, the rationale for the narrowing amendment was to distinguish the claimed coherent detection module from Takahashi's WDM multiplexer/demultiplexer architecture with an integrated coherent receiver. The applicant's September 16, 2019 response argued that Takahashi's coherent receiver was structurally distinct because it was part of a WDM multiplexing/demultiplexing scheme and lacked the "unitary" module configuration recited in the amended claims. The core of the argument for patentability over Takahashi was that the claimed invention provided a self-contained, unitary coherent detection module distinct from Takahashi's distributed WDM receiver architecture.
The equivalent now at issue — Lumicore's distributed coherent detection architecture — is not the same type of design as Takahashi's. Takahashi's reference describes a fundamentally different system: a dense WDM silicon photonic multiplexer/demultiplexer with coherent receiver functionality integrated into a multi-channel WDM architecture. Lumicore's accused products, by contrast, are single-channel (or multi-lane, non-WDM) coherent transceivers that employ a distributed layout of coherent detection components on a silicon photonic die. The reason Lumicore's products do not literally infringe — the distributed (rather than unitary) arrangement of the optical hybrid, photodetectors, and TIA — is a different structural variation than the one that motivated the narrowing amendment, which was focused on distinguishing the claimed invention from Takahashi's multi-channel WDM receiver architecture.
In other words, the amendment was directed at distinguishing the type of system (unitary coherent module vs. WDM receiver architecture) rather than the specific spatial arrangement of coherent detection components that makes Lumicore's design different from the literal claim language. The rationale for the amendment — addressing Takahashi's WDM-based coherent receiver — bore only a tangential relation to the specific equivalent at issue — Lumicore's distributed single-channel coherent detection architecture.
The Federal Circuit's decision in Insituform Technologies, Inc. v. CAT Contracting, Inc., 385 F.3d 1360, 1370 (Fed. Cir. 2004), supports this argument. In Insituform, the court held that the tangential relation exception applied where the amendment was made to address a prior-art reference that was structurally different from the accused equivalent, even though the amendment narrowed the same claim limitation. The court emphasized that the inquiry focuses on "the relationship between the reason for the amendment and the alleged equivalent," not merely on whether the same limitation was amended. Id.
This is our strongest ground for overcoming estoppel, and we recommend that it be the lead argument in the P.R. 3-1 Infringement Contentions, in Dr. Ashford's expert report, and in any opposition to summary judgment.
##### b. Unforeseeability Exception (Secondary Argument)
As a secondary argument, we assess whether Lumicore's specific modified coherent detection architecture was unforeseeable at the time of the narrowing amendment (September 16, 2019). Under the unforeseeability exception, the Festo presumption is rebutted if "the equivalent may have been unforeseeable at the time of the amendment." Festo, 535 U.S. at 740.
The Federal Circuit has interpreted this exception to apply most readily to "later-developed technologies" that were not known or available at the time of prosecution. Festo, 344 F.3d at 1369 (on remand). The exception is more difficult to establish where the equivalent is based on technology that was known in the art at the time of the amendment.
There is a plausible argument that Lumicore's specific distributed coherent detection architecture, with its particular MMI coupler network geometry and flip-chip TIA integration approach, was not foreseeable in September 2019. Lumicore's design documents indicate that the specific distributed architecture was developed internally at Lumicore between 2020 and 2021 (Bates Nos. LUMI-00042187 through LUMI-00042203), and the earliest published description of a comparable architecture in the academic literature appears to date from late 2020. However, the general concept of distributing coherent detection components across a photonic die (rather than integrating them in a compact unitary module) was known in the silicon photonics research community as early as 2017–2018. Several academic papers from this period describe distributed coherent receiver architectures on silicon photonic platforms, although none employ the specific design used by Lumicore.
This distinction — between the general concept of a distributed architecture (known at the time) and the specific design implemented by Lumicore (arguably not foreseeable) — presents a factual question that requires further development through expert analysis. We recommend that Dr. Ashford address this issue in her expert report, with a thorough survey of the state of the art as of September 2019 to determine whether the specific structural variation employed by Lumicore was foreseeable at that time.
Assessment: This argument is viable but weaker than the tangential-relation exception. If the general concept of distributed coherent detection was known in the art at the time of the amendment, Lumicore will argue that the specific variation was foreseeable, even if it had not yet been implemented in a commercial product. We recommend developing this argument fully but presenting it as a secondary basis for overcoming estoppel, not as the primary ground.
##### c. Other Reason Exception
The third Festo exception — that "there [is] some other reason suggesting that the patentee could not reasonably be expected to have described the insubstantial substitute in question," Festo, 535 U.S. at 740 — is narrowly applied by the Federal Circuit and is generally considered a "safety valve" for cases that do not fit neatly into the unforeseeability or tangential-relation categories. See Festo, 344 F.3d at 1369.
We have identified one potential basis for this exception: the argument that the specific structural difference between the claimed unitary module and the accused distributed architecture is a difference of degree (spatial layout) rather than of kind (detection mechanism), and that it would have been unreasonable to expect the applicant to have described every possible spatial layout of the coherent detection components during prosecution. The claim already describes the functional elements (optical hybrid, balanced photodetectors, TIA); the narrowing amendment focused on the "unitary" integration of these elements, and it would have been impracticable for the applicant to catalogue all possible non-unitary spatial distributions as equivalents.
Assessment: This argument has some force but is unlikely to succeed on its own. The Federal Circuit has generally been skeptical of arguments that the applicant could not have drafted broader claim language to capture the equivalent. We recommend presenting this argument only as a supplement to the tangential-relation and unforeseeability exceptions, not as a standalone basis for overcoming estoppel.
3. Overall Estoppel Risk Assessment for '193 Claims 3 and 7. Risk Level: Moderate.
The Festo presumption clearly applies because the "coherent detection module" limitation was narrowed during prosecution to overcome a § 102 rejection. The tangential-relation exception is viable and well-supported: the amendment was directed at distinguishing over Takahashi's WDM-based coherent receiver architecture, which is a structurally different design from Lumicore's distributed single-channel coherent detection architecture. The unforeseeability exception provides secondary support, though its strength depends on further factual development regarding the state of the art as of September 2019.
Defense counsel Kate Rourke at Hawthorne Vega & Cruz LLP will argue that the amendment directly targeted the "coherent detection module" limitation — the very limitation on which OWS now asserts DOE — and that estoppel should therefore foreclose DOE on that limitation. This is a legitimate argument, and the Court could find it persuasive. However, the Festo inquiry does not simply ask whether the same limitation was amended; it asks whether the rationale for the amendment bears on the specific equivalent at issue. The Supreme Court in Festo made clear that "the patentee's decision to narrow his claims through amendment may not be presumed to be an admission that the territory surrendered in an amendment is fair game for all purposes of patent law." 535 U.S. at 737.
We recommend proceeding with DOE contentions on Claims 3 and 7, leading with the tangential-relation argument. Dr. Ashford's expert report should clearly articulate why the amendment was directed at a different aspect of the prior art (Takahashi's WDM architecture) than the equivalent at issue (Lumicore's distributed coherent detection layout). We should also prepare a contingency opposition brief to Lumicore's anticipated summary judgment motion on DOE/prosecution history estoppel, which we expect to be filed in or around September 2025 in advance of the dispositive motions deadline.
IV. Analysis of '015 Patent — Claim 12
A. Claim Language
The relevant claims of the '015 Patent are set forth in full in Appendix D. Claim 12 is a dependent claim. For purposes of this analysis, we reproduce the pertinent claim language below.
Claim 9 (independent, parent claim) recites:
9. A polarization-diverse grating coupler for a silicon photonic transceiver, comprising:
a silicon waveguide layer disposed on a buried oxide layer of a silicon-on-insulator substrate;
a grating structure patterned in the silicon waveguide layer and configured to receive an optical signal from an optical fiber positioned above the grating structure;
a first output waveguide configured to carry a first polarization component of the optical signal; and
a second output waveguide configured to carry a second polarization component of the optical signal, the second polarization component being orthogonal to the first polarization component,
wherein the grating structure is configured to simultaneously couple and separate the first polarization component and the second polarization component from the optical signal into the first output waveguide and the second output waveguide, respectively.
Claim 10 (dependent on Claim 9) recites:
10. The polarization-diverse grating coupler of claim 9, wherein the grating structure comprises a periodic pattern of scattering elements configured to diffract the optical signal into the silicon waveguide layer.
Claim 12 (dependent on Claim 10, and through it on Claim 9) recites:
12. The polarization-diverse grating coupler of claim 10, wherein the grating structure comprises a two-dimensional sub-wavelength grating having a first periodic pattern along a first axis and a second periodic pattern along a second axis perpendicular to the first axis, the first periodic pattern having a first period selected to couple the first polarization component, and the second periodic pattern having a second period selected to couple the second polarization component, wherein the first period and the second period are each less than an operating wavelength of the optical signal divided by an effective refractive index of the silicon waveguide layer.
The "two-dimensional sub-wavelength grating" limitation is the focus of the DOE analysis for Claim 12. The '015 Patent specification describes this structure in detail at column 11, lines 4–48, and column 12, lines 1–22. The specification explains that the two-dimensional sub-wavelength grating "enables simultaneous coupling and polarization splitting in a single grating element by exploiting the anisotropic diffraction properties of a two-dimensional periodic structure with sub-wavelength feature sizes." '015 Patent, col. 11, ll. 8–12. The specification further states that "the sub-wavelength nature of the grating periods ensures that no higher-order diffraction modes propagate in the waveguide layer, thereby achieving high coupling efficiency and low polarization-dependent loss." Id., col. 11, ll. 34–38.
B. Prosecution History of the '015 Patent — The "Two-Dimensional Sub-Wavelength Grating" Amendment
The prosecution history of the '015 Patent is directly relevant to the DOE analysis because the "two-dimensional sub-wavelength grating" limitation was added to the claims by amendment during prosecution. The relevant prosecution history excerpts are included in Appendix B; we summarize the key events below.
On February 4, 2021, the PTO issued a non-final Office Action rejecting the original claims of the '015 Patent under 35 U.S.C. § 103 as obvious over a combination of two references: U.S. Patent No. 8,412,007 (Bergmann, "Polarization-Independent Grating Coupler for Silicon Waveguides," issued April 2, 2013) and International Patent Application PCT/EP2015/071234 (Simonsen, "Integrated Optical Coupler with Polarization Diversity," filed September 17, 2015). The examiner found that Bergmann disclosed a grating coupler with a periodic scattering pattern for coupling light from a fiber into a silicon waveguide, and that Simonsen disclosed a polarization-diverse coupling approach with separate output waveguides for orthogonal polarization states. The examiner concluded that it would have been obvious to combine the teachings of Bergmann and Simonsen to arrive at the claimed polarization-diverse grating coupler.
Importantly, Bergmann's grating coupler employed a one-dimensional grating structure — a periodic pattern along a single axis — that did not inherently provide polarization splitting. Simonsen's coupler achieved polarization diversity through a different geometry — a tapered waveguide structure downstream of a conventional grating coupler, rather than through a two-dimensional grating with inherent polarization-splitting properties.
On August 12, 2021, the applicant (through Dr. Martin Chu at Cascade IP Advisors) filed a response to the Office Action that included an amendment to the dependent claim that became Claim 12. The amendment added the "two-dimensional sub-wavelength grating" modifier to the grating limitation. The original claim language covered grating couplers with a periodic pattern of scattering elements generally, without specifying the two-dimensional geometry or the sub-wavelength dimensional constraint. The amendment introduced the requirement of (1) a two-dimensional grating with periodic patterns along two perpendicular axes, (2) with each period selected to couple a respective polarization component, and (3) with each period being less than the operating wavelength divided by the effective refractive index (i.e., sub-wavelength).
In the accompanying remarks, Dr. Chu argued that the amended claim language distinguished over Bergmann and Simonsen because:
"Neither Bergmann nor Simonsen, alone or in combination, teaches or suggests a two-dimensional sub-wavelength grating with the specific structural and functional characteristics recited in amended Claim 12. Bergmann discloses only a one-dimensional grating coupler without polarization-splitting capability, and Simonsen achieves polarization diversity through a tapered waveguide structure downstream of the grating, not through the grating itself. The two-dimensional sub-wavelength grating of amended Claim 12 achieves simultaneous coupling and polarization splitting in a single grating element by virtue of its two-dimensional periodic structure with sub-wavelength feature sizes — a feature absent from both references."
The examiner allowed the claims. A Notice of Allowance was issued on March 22, 2022, and the patent issued on August 2, 2022.
The narrowing amendment — adding the "two-dimensional sub-wavelength" modifier — was made to overcome the § 103 obviousness rejection based on Bergmann and Simonsen. This was an amendment for reasons of patentability, triggering the Festo presumption of prosecution history estoppel. The scope of surrender encompasses the territory between the original broader grating coupler limitation (which covered any grating coupler with a periodic pattern of scattering elements) and the narrowed limitation requiring a "two-dimensional sub-wavelength grating" with the specific structural characteristics described in Claim 12.
C. Lumicore's Accused Design — How It Differs from the Literal Claim
The Benchmark teardown reports for the LumiLink 800G-DR8 (June 2023) and the LumiEdge 1600G-LR (February 2024) reveal that Lumicore's accused products employ a grating coupler structure that achieves polarization-diverse coupling of light from a fiber into the silicon photonic waveguide layer. However, the accused grating structure may not literally satisfy the "two-dimensional sub-wavelength grating" limitation of Claim 12.
Specifically, Benchmark's analysis — based on scanning electron microscopy (SEM) imaging and focused ion beam (FIB) cross-sectioning of the accused grating structures — indicates that Lumicore's grating coupler employs a quasi-periodic two-dimensional grating geometry in which the grating periods along one of the two axes slightly exceed the sub-wavelength threshold defined in the claim (i.e., the period along one axis is slightly greater than the operating wavelength divided by the effective refractive index of the silicon waveguide layer). The grating is two-dimensional in the sense that it has periodic structures along two perpendicular axes, but the dimensional constraint of "sub-wavelength" (as defined in the claim) may not be strictly satisfied along one axis.
Kevin Ng's Benchmark report for the LumiLink 800G-DR8 states:
"SEM imaging of the grating coupler die region reveals a two-dimensional grating pattern with nominal periods of approximately 580 nm along the x-axis and approximately 640 nm along the y-axis. Given the operating wavelength of 1310 nm and the estimated effective refractive index of the SOI waveguide layer of approximately 2.1, the sub-wavelength threshold is approximately 624 nm (1310 nm / 2.1). The x-axis period of 580 nm is below this threshold, but the y-axis period of 640 nm slightly exceeds it."
Benchmark Teardown Report, LumiLink 800G-DR8, at 51. The LumiEdge 1600G-LR exhibits a similar grating structure with comparable dimensional parameters. See Benchmark Teardown Report, LumiEdge 1600G-LR, at 33.
The fact that the y-axis period (approximately 640 nm) slightly exceeds the sub-wavelength threshold (approximately 624 nm) means that Lumicore's accused grating may not literally satisfy the claim requirement that "the first period and the second period are each less than an operating wavelength of the optical signal divided by an effective refractive index of the silicon waveguide layer." This is the specific structural difference that motivates the DOE analysis for Claim 12.
Despite this dimensional departure, the accused grating structure accomplishes the same polarization-diverse coupling function as the claimed two-dimensional sub-wavelength grating. The grating achieves simultaneous coupling and polarization splitting of both TE and TM modes from the fiber into separate output waveguides. The slight exceedance of the sub-wavelength threshold along one axis does not prevent the grating from operating as a polarization-diverse coupler; it merely means that a low-power higher-order diffraction mode may propagate along that axis, which Lumicore suppresses through additional design features (apodization and absorptive boundary structures at the waveguide edges, per Lumicore design files at Bates No. LUMI-00078334).
D. Function-Way-Result Analysis
We apply the function-way-result test to the "two-dimensional sub-wavelength grating" limitation of Claim 12 as applied to Lumicore's accused products.
1. Function. The claimed "two-dimensional sub-wavelength grating" functions as a polarization-splitting coupling element that simultaneously couples both TE and TM polarization modes from an optical fiber into the silicon photonic integrated circuit, directing the two orthogonal polarization components into separate output waveguides. This dual-function coupling-and-splitting operation is achieved in a single grating element, eliminating the need for a separate polarization beam splitter downstream of the grating coupler.
Lumicore's accused grating coupler performs substantially the same function. The accused two-dimensional grating structure simultaneously couples both TE and TM polarization modes from the input fiber into the

PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT

MEMORANDUM

Record

Parties

Dates

Amounts