FR EA
1014-00003 · Optiwave Semiconductor Corp · Post-Markman Assessment
post-markman-viability-678-patent.docx DOCUMENT Litigation
63 / 83 ← → ✕
Contract analysis Extract to table

PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT This memorandum constitutes attorney work product prepared in anticipation of litigation and is protected from disclosure under Fed. R. Civ. P. 26(b)(3) and the work product doctrine. It also contains confidential attorney-client communications protected by the attorney-client privilege. This memorandum should not be disclosed to any person or entity outside of Calderwood & Harkness LLP and authorized representatives of Optiwave Semiconductor Corp. without the express prior written consent of counsel.

MEMORANDUM

TO: Alan Ngo, Responsible Partner, intellectual Property Group, Calderwood & Harkness LLP

FROM: Omar Siddiqui, Lead Associate, intellectual Property Group, Calderwood & Harkness LLP

DATE: February 22, 2023

RE: Post-Markman Infringement Viability Analysis — U.S. Patent No. 11,045,678 ("the '678 Patent") — Optiwave Semiconductor Corp. v. Lumos Photonics, Inc., Case No. 5:22-cv-03417-EJD (N.D. Cal.)

MATTER NO.: 1014-00003

Right-click to update Table of Contents

I. Executive Summary

The purpose of this memorandum is to assess the continued viability of Optiwave Semiconductor Corp.'s ("OWS") infringement theories with respect to U.S. Patent No. 11,045,678 ("the '678 Patent") following the Markman Order issued on February 17, 2023, by the Honorable Edward J. Torvelis in the United States District Court for the Northern District of California. This analysis is intended to inform the litigation team's post-Markman strategy and to support the comprehensive settlement evaluation currently underway.

The Court's construction of the critical claim term — "integrated wavelength-division multiplexing element monolithically formed on a single silicon die" — adopted a construction substantially aligned with Defendant Lumos Photonics, Inc.'s ("Lumos") proposed construction. Specifically, the Court construed the term to mean: "fabricated entirely within the same front-end-of-line and back-end-of-line semiconductor processing sequence on a single silicon substrate without heterogeneous bonding or attachment of separately fabricated optical components." This construction is materially narrower than OWS's proposed construction, which defined the term as "a WDM component fabricated as part of the same semiconductor fabrication process on a single die."

The practical impact of the Court's construction is severe. The construction effectively excludes Lumos's LumosBridge™ 800G Optical Engine and LumosEdge™ 400ZR+ Pluggable Transceiver platforms from the scope of the asserted claims. Both accused products employ micro-transfer-printed III-V (indium phosphide-based) gain elements that are separately fabricated and subsequently bonded to the silicon photonic integrated circuit ("PIC"). This heterogeneous integration technique falls squarely within the exclusionary language of the Court's construction — specifically, the prohibition on "heterogeneous bonding or attachment of separately fabricated optical components."

The disputed claim term appears in independent Claim 1 and is incorporated by reference into all four dependent asserted claims (Claims 2, 6, 9, and 14). Because Lumos's products do not satisfy the "monolithically formed" limitation as construed by the Court, literal infringement of all five asserted claims is untenable.

Furthermore, the doctrine of equivalents does not provide a viable alternative path to infringement. The prosecution history of the '678 Patent conclusively establishes prosecution history estoppel under Festo Corp. v. Fenmore & Co., 535 U.S. 722 (2002). During prosecution, OWS narrowed the claims by adding the "monolithically formed" limitation specifically to distinguish prior art references that disclosed hybrid and heterogeneous integration techniques. The applicant's own remarks during prosecution stated unequivocally that "the present invention is distinguished from heterogeneous integration approaches." This clear and deliberate narrowing amendment forecloses any attempt to recapture the surrendered claim scope through the doctrine of equivalents. None of the three Festo rebuttal exceptions apply.

Bottom-Line Conclusion: Infringement of all five asserted claims of the '678 Patent (Claims 1, 2, 6, 9, and 14) is untenable under the Court's claim construction. The doctrine of equivalents is foreclosed by prosecution history estoppel. This patent should not be relied upon as a basis for continuing litigation against Lumos. This conclusion should be incorporated into the overall post-Markman settlement analysis and the Board presentation scheduled for February 24, 2023.

II. Background

A. The '678 Patent Overview

U.S. Patent No. 11,045,678 is titled "Multi-Channel Optical Transceiver with Integrated Wavelength-Division Multiplexing on Silicon." The patent application was filed on June 22, 2018, and issued on June 29, 2021. The named inventors are Dr. Priya Ramanathan (OWS Chief Executive Officer and Co-Founder), Dr. Amara Osei (OWS Principal Engineer), and Liwei Zhang (OWS Design Engineer). The '678 Patent is assigned to OWS.

The '678 Patent is generally directed to a multi-channel optical transceiver architecture that integrates wavelength-division multiplexing ("WDM") elements monolithically on a single silicon die. The patent teaches that by monolithically forming the WDM multiplexer and demultiplexer elements — such as arrayed waveguide gratings ("AWGs"), echelle gratings, or ring resonator-based structures — within the same silicon fabrication process used to create the transceiver's waveguides, modulators, and photodetectors, the resulting optical transceiver can achieve compact form factors, reduced assembly costs, improved optical performance, and high-bandwidth optical interconnects suitable for data center and telecommunications applications. The specification describes embodiments achieving aggregate data rates of 400 Gbps and above through the use of multiple wavelength channels carried on a single optical fiber, with the WDM elements performing the channel multiplexing and demultiplexing functions on-chip.

The patent contains 18 claims in total, of which Claims 1 and 9 are independent claims. However, in this litigation, OWS asserted only Claims 1, 2, 6, 9, and 14 against Lumos. Upon closer review of the claim structure, Claim 9 is a dependent claim that depends from Claim 1, notwithstanding its numbering. All five asserted claims therefore trace back to independent Claim 1 and incorporate its limitations, including the critical "monolithically formed" limitation at issue.

OWS's own key products that practice the '678 Patent include the OWS-400ZR and OWS-800LR coherent optical transceiver chipsets. These chipsets are fabricated by Pinnacle Semiconductor Foundry Co., Ltd. on a 45nm silicon-on-insulator ("SOI") process. In OWS's products, the WDM elements, silicon waveguides, germanium photodetectors, and silicon-based modulators are all fabricated within the same monolithic SOI process — the design approach that the '678 Patent claims to protect. OWS does not employ micro-transfer printing, flip-chip bonding, or any other heterogeneous integration technique for the optical components in these particular products.

B. Accused Products

OWS has accused two Lumos products of infringing the '678 Patent:

LumosBridge™ 800G Optical Engine. The LumosBridge 800G is a co-packaged optics engine designed for deployment in hyperscale data center switches. It provides 800 Gbps aggregate bandwidth through eight wavelength channels at 100 Gbps per channel, utilizing a dense WDM architecture. The product was first shipped in Q1 2022 and is sold at a unit price of approximately $1,200 to $1,800, depending on configuration and volume. The LumosBridge 800G is designed to be integrated directly into switch ASICs, eliminating the need for traditional pluggable optical modules and reducing power consumption and latency.

LumosEdge™ 400ZR+ Pluggable Transceiver. The LumosEdge 400ZR+ is a pluggable coherent optical transceiver in a QSFP-DD form factor. It is designed for metro and data center interconnect ("DCI") applications, providing 400 Gbps coherent transmission over distances of up to 120 kilometers. The LumosEdge 400ZR+ has been in production since Q3 2021 and is sold at a unit price of approximately $600 to $900.

According to Lumos's 10-K filing for the fiscal year ended December 31, 2022, and corroborated by Lumos's responses to OWS's interrogatories and requests for production (served August 15, 2022), the combined revenue attributable to the accused products for calendar year 2022 was approximately $148.6 million. This figure encompasses both domestic and international sales, though OWS's infringement claims are limited to domestic sales and importation under 35 U.S.C. §§ 271(a) and (b).

The key technical feature at issue in this litigation is the architecture of Lumos's silicon photonic integrated circuit. Both accused products utilize a silicon PIC that includes monolithically formed silicon waveguides, silicon-based ring resonator modulators, germanium-on-silicon photodetectors, and WDM multiplexer/demultiplexer elements (specifically, ring resonator-based MUX/DEMUX structures in the LumosBridge 800G and echelle grating-based MUX/DEMUX structures in the LumosEdge 400ZR+). These WDM elements are indeed fabricated on the silicon substrate within the standard front-end-of-line and back-end-of-line processing sequence.

However, critically, both accused products also incorporate micro-transfer-printed III-V gain elements — specifically, indium phosphide (InP)-based distributed feedback ("DFB") lasers and semiconductor optical amplifiers ("SOAs") — that are separately fabricated on an InP substrate and then transferred to the silicon PIC using a micro-transfer printing process. In micro-transfer printing, the III-V gain elements are first fabricated on their native InP substrate, then released from that substrate using selective etching, picked up by an elastomeric stamp, and precisely placed and bonded onto designated receptor sites on the silicon PIC. This process constitutes a heterogeneous integration technique — the III-V gain elements are "separately fabricated optical components" that are "attached" to the silicon substrate through a bonding process, rather than being fabricated within the same semiconductor processing sequence as the silicon-based components.

This architectural distinction — monolithic silicon photonics versus heterogeneous integration of III-V gain elements — is the central technical dispute in this case and the issue upon which the Markman ruling was dispositive.

C. Pre-Markman Infringement Theory (Summary)

OWS's pre-suit infringement analysis was memorialized in a detailed memorandum dated May 5, 2022, prepared by Christy Bauer, then a senior associate in Calderwood & Harkness's intellectual Property Group. That memorandum provided the foundation for OWS's decision to file suit and identified the infringement theories for each of the four asserted patents.

With respect to the '678 Patent, the pre-suit memorandum identified the "monolithically formed on a single silicon die" limitation as a potential vulnerability in the infringement case. Ms. Bauer correctly recognized that the viability of the infringement theory depended critically on claim construction. The memorandum articulated OWS's proposed construction — "a WDM component fabricated as part of the same semiconductor fabrication process on a single die" — and explained that under this construction, the focus would be on whether the WDM element itself was fabricated on the silicon die, not on whether every optical component on the die was fabricated in the same monolithic process.

Under OWS's proposed construction, the infringement theory was tenable. Lumos's WDM elements (ring resonator MUX/DEMUX structures and echelle gratings) are unquestionably fabricated on the silicon substrate within the standard SOI fabrication process. If the "monolithically formed" limitation were construed to apply only to the WDM element — as OWS proposed — then the presence of separately fabricated and bonded III-V gain elements elsewhere on the die would be irrelevant to the infringement analysis.

However, the pre-suit memorandum also flagged the risk that the Court might adopt a narrower construction — one that would require the entire optical transceiver on the die, or at least all optical components on the die, to be monolithically fabricated without any heterogeneous integration. Under such a construction, Ms. Bauer noted, Lumos's micro-transfer-printed III-V architecture would "likely fall outside the claim scope." The memorandum rated the infringement probability on the '678 Patent as "moderate risk" and identified claim construction as "the pivotal issue" that would determine the outcome.

The Markman Order has now resolved this pivotal issue against OWS, confirming the worst-case scenario that the pre-suit analysis anticipated.

III. The Markman Order — Construction of Disputed Terms for the '678 Patent

A. The Key Disputed Term

The central disputed claim term for the '678 Patent was: "integrated wavelength-division multiplexing element monolithically formed on a single silicon die." This term appears in independent Claim 1 at lines 4–6 of the claim body and is incorporated by reference into all four dependent asserted claims — Claims 2, 6, 9, and 14 — each of which depends directly from Claim 1. The construction of this single term is therefore dispositive of the infringement analysis for all five asserted claims.

The parties submitted competing proposed constructions for this term as part of their Joint Claim Construction Statement filed on November 18, 2022. The parties' technical tutorials were presented to the Court on December 16, 2022, and the Markman hearing was held on January 27, 2023. The Court's Markman Order was issued on February 17, 2023, addressing all 11 disputed terms across the four asserted patents.

B. OWS's Proposed Construction

OWS proposed that the term "integrated wavelength-division multiplexing element monolithically formed on a single silicon die" be construed to mean: "a WDM component fabricated as part of the same semiconductor fabrication process on a single die."

OWS's construction was anchored in three principal arguments. First, OWS argued that the modifier "monolithically formed" grammatically and structurally modifies "wavelength-division multiplexing element" — i.e., it is the WDM element that must be monolithically formed, not the entire optical transceiver or every component on the die. Under this reading, the claim requires only that the WDM element be fabricated using the monolithic silicon process; other optical components on the die (such as gain elements) could be integrated using different techniques without affecting the "monolithically formed" limitation.

Second, OWS pointed to several passages in the specification that described embodiments in which non-WDM components were integrated using techniques other than monolithic fabrication. For example, the specification at column 8, lines 34–42, describes an embodiment in which "external gain elements may be coupled to the silicon die to provide optical amplification." OWS argued that these embodiments demonstrated that the specification contemplated configurations where not all optical components were monolithically formed, and that the claims should be construed consistently with the full scope of the disclosed embodiments.

Third, OWS argued that its broader construction was consistent with the ordinary meaning of "monolithically formed" in the silicon photonics field, which OWS's expert, Dr. Helena Voss of the Massachusetts Institute of Technology, opined referred to the fabrication of a specific component using the same process steps as the underlying silicon substrate, without requiring that every other component on the die also be monolithically fabricated.

C. Lumos's Proposed Construction

Lumos proposed that the term be construed to mean: "a WDM component formed during front-end-of-line processing within the same silicon substrate, excluding hybrid integration, flip-chip bonding, or any heterogeneous integration technique."

Lumos's construction was grounded primarily in the prosecution history of the '678 Patent. Lumos argued that the term "monolithically formed on a single silicon die" was added by amendment during prosecution specifically to overcome prior art rejections based on references that disclosed heterogeneous integration techniques, and that the applicant's prosecution remarks constituted a clear and unambiguous disavowal of claim scope encompassing heterogeneous integration.

Lumos further argued that the intrinsic record as a whole — including the specification's emphasis on a "single, unified fabrication process" (column 5, lines 12–18), the prosecution history amendments, and the absence of any claim language or specification language suggesting that "monolithically formed" could include separately fabricated and bonded components — required a construction that excluded heterogeneous integration techniques in their entirety.

Lumos's expert, Dr. Karl-Friedrich Brenner of Stanford University, opined that one of ordinary skill in the art of silicon photonics would understand "monolithically formed on a single silicon die" to mean fabrication within a single, unified semiconductor processing sequence, inherently excluding any process in which separately fabricated components are bonded, transferred, or otherwise attached to the silicon substrate.

D. The Court's Construction

After considering the intrinsic evidence — including the claim language, specification, and prosecution history — as well as the extrinsic evidence presented through expert testimony and technical literature, the Court construed the disputed term as follows:

> "Fabricated entirely within the same front-end-of-line and back-end-of-line semiconductor processing sequence on a single silicon substrate without heterogeneous bonding or attachment of separately fabricated optical components."

The Court's construction largely adopted Lumos's position, with a minor but notable expansion: the Court included reference to back-end-of-line ("BEOL") processing in addition to front-end-of-line ("FEOL") processing, recognizing that certain optical components (such as waveguide cladding layers and metallization for photodetector contacts) are formed during BEOL steps as part of the same unified fabrication sequence. Lumos's proposed construction had referenced only FEOL processing, which the Court found to be unduly narrow in this one respect. However, the Court critically retained the core exclusionary language — "without heterogeneous bonding or attachment of separately fabricated optical components" — that is fatal to OWS's infringement theories.

The Court's analysis of this term spans pages 18 through 23 of the 34-page Markman Order and addresses the intrinsic evidence in detail.

Claim Language. The Court began by analyzing the claim language itself, noting that the phrase "monolithically formed on a single silicon die" is a participial phrase that modifies "wavelength-division multiplexing element" but does so in the context of an "integrated" optical transceiver. The Court found that the term "monolithically," derived from the Greek monolithos (meaning "single stone"), carries an inherent connotation of unity and indivisibility in semiconductor fabrication. The Court concluded that "monolithically formed on a single silicon die" would be understood by one of ordinary skill in the art to require fabrication within a single, unified semiconductor process, without the introduction of separately fabricated components through bonding or transfer techniques.

Specification. The Court examined the specification and found that the preferred embodiment language consistently emphasized a "single, unified fabrication process" (col. 5, ll. 12–18) and described the advantages of monolithic integration in terms of eliminating the need for "separate alignment, bonding, and packaging steps associated with discrete component assembly" (col. 3, ll. 45–52). The Court acknowledged OWS's argument regarding the embodiment at column 8, lines 34–42, which references "external gain elements," but found that this passage described an alternative embodiment that was not claimed and did not expand the meaning of "monolithically formed" as used in the claims. The Court noted that "[a] patentee's decision to describe but not claim certain embodiments does not broaden the scope of the claims that were actually drafted." Markman Order at 20.

Prosecution History. The Court placed the greatest weight on the prosecution history, which it characterized as "unambiguous and dispositive." Id. at 21. The Court traced the prosecution of the '678 Patent through three office action rejections:

• In the first office action, the examiner rejected the original claims under 35 U.S.C. § 102(a)(1) over U.S. Patent No. 9,612,401 to Chen et al. ("Chen"), which disclosed a multi-channel optical transceiver with heterogeneously integrated III-V gain elements on a silicon photonic platform. The original claims recited only "an optical transceiver integrated on a single die" without the "monolithically formed" limitation.

• In the second office action (after the applicant's first response), the examiner maintained the rejection, finding that the applicant's arguments did not adequately distinguish the claimed invention from Chen's heterogeneously integrated transceiver. The examiner noted that the distinction between monolithic and heterogeneous integration was "not clearly reflected in the claim language."

• In response to the third office action, the applicant amended Claim 1 to add the "monolithically formed" limitation and submitted remarks stating, in pertinent part: "The present invention is distinguished from heterogeneous integration approaches such as those disclosed in Chen. The claimed multi-channel optical transceiver requires a WDM element that is monolithically formed on a single silicon die, meaning fabricated within the same semiconductor processing sequence without the need for separate fabrication and bonding of optical components from different material systems. Chen's transceiver, which relies on heterogeneously bonded III-V gain elements, does not meet this limitation."

The Court found that these prosecution remarks constituted "a clear and unambiguous disavowal of claim scope encompassing heterogeneous integration techniques." Markman Order at 22. The Court emphasized that the applicant did not merely argue that the prior art failed to disclose the claimed invention; rather, the applicant "expressly defined the boundary of the invention by reference to the exclusion of heterogeneous integration." Id. The Court held that "construing 'monolithically formed' to encompass heterogeneous bonding would be inconsistent with the narrowing amendments made to secure allowance and would effectively recapture the claim scope that the applicant surrendered during prosecution." Id. at 23.

Extrinsic Evidence. The Court considered but did not rely heavily on the competing expert testimony, finding that the intrinsic record was sufficiently clear to resolve the dispute. The Court noted, however, that both experts agreed that "monolithically formed" in the silicon photonics field generally refers to fabrication within a single process sequence, and that the dispute was primarily about whether the limitation applied to the WDM element alone or to the die as a whole. The Court resolved this dispute in Lumos's favor based on the prosecution history, which demonstrated that the applicant intended the limitation to exclude any die incorporating heterogeneously bonded optical components.

E. Additional Disputed Terms for the '678 Patent

Of the 11 total disputed claim terms across all four asserted patents, the term discussed above — "integrated wavelength-division multiplexing element monolithically formed on a single silicon die" — was the only term from the '678 Patent that was the subject of extensive briefing and argument. One additional term from the '678 Patent, "optically coupled to a common output waveguide," was included in the Joint Claim Construction Statement but was resolved by stipulation of the parties prior to the Markman hearing. The parties agreed that this term carries its plain and ordinary meaning: "connected such that light is transmitted from one optical component to another through a shared output waveguide." This stipulated construction does not affect the infringement analysis, as the dispositive issue remains the "monolithically formed" limitation.

No other terms from the '678 Patent were disputed. The remaining 9 of the 11 total disputed terms pertained to the '567 Patent, the '234 Patent, and the '890 Patent, the analyses of which are addressed in separate companion memoranda.

IV. Claim-by-Claim Infringement Viability Analysis

A. Claim 1 (Independent Claim)

1. Claim Language

Claim 1 of the '678 Patent reads as follows:

1\. A multi-channel optical transceiver, comprising:

a silicon photonic integrated circuit formed on a silicon-on-insulator substrate;

an integrated wavelength-division multiplexing element monolithically formed on a single silicon die, the wavelength-division multiplexing element comprising at least one of an arrayed waveguide grating, an echelle grating, or a plurality of ring resonators configured to multiplex or demultiplex a plurality of optical channels at distinct wavelengths;

a plurality of optical modulators, each optical modulator configured to modulate an optical carrier signal at a respective one of the distinct wavelengths;

a plurality of photodetectors, each photodetector configured to detect an optical signal at a respective one of the distinct wavelengths;

a common optical interface configured to couple the multiplexed optical channels to an optical fiber; and

electronic control circuitry operably connected to the plurality of optical modulators and the plurality of photodetectors,

wherein the wavelength-division multiplexing element, the plurality of optical modulators, the plurality of photodetectors, and the common optical interface are co-integrated on the single silicon die.

The critical limitation for purposes of this analysis is: "an integrated wavelength-division multiplexing element monolithically formed on a single silicon die." This limitation, as construed by the Court, requires that the WDM element be "fabricated entirely within the same front-end-of-line and back-end-of-line semiconductor processing sequence on a single silicon substrate without heterogeneous bonding or attachment of separately fabricated optical components."

The "wherein" clause at the end of Claim 1 further requires that the WDM element, modulators, photodetectors, and common optical interface be "co-integrated on the single silicon die." While this clause reinforces the integration requirement, the dispositive issue under the Court's construction is whether the die satisfies the "monolithically formed" limitation, which the Court has construed to exclude dies incorporating any heterogeneously bonded or separately fabricated optical components.

2. Application of the Court's Construction to Lumos's Products

Under the Court's construction, the question is whether Lumos's accused products — the LumosBridge 800G and LumosEdge 400ZR+ — are "fabricated entirely within the same front-end-of-line and back-end-of-line semiconductor processing sequence on a single silicon substrate without heterogeneous bonding or attachment of separately fabricated optical components."

The answer is unambiguously no. The following analysis applies to both accused products.

Micro-Transfer-Printed III-V Gain Elements. Both the LumosBridge 800G and the LumosEdge 400ZR+ incorporate micro-transfer-printed III-V gain elements on their respective silicon PICs. Specifically, the LumosBridge 800G includes eight InP-based DFB laser gain elements and four InP-based SOAs, all of which are separately fabricated on an InP substrate at Lumos's III-V fabrication partner, Epitech Compound Semiconductor GmbH, and subsequently transferred to the silicon PIC via micro-transfer printing. The LumosEdge 400ZR+ similarly includes four InP-based DFB laser gain elements and two InP-based SOAs, fabricated and transferred using the same process.

These III-V gain elements are, by any definition, "separately fabricated optical components" that are "attached" to the silicon substrate through a heterogeneous bonding process. The micro-transfer printing technique involves: (1) fabricating the III-V gain elements on their native InP substrate using standard III-V epitaxial growth and processing; (2) selectively etching a release layer beneath the gain elements to partially free them from the InP substrate; (3) using an elastomeric polydimethylsiloxane (PDMS) stamp to pick up the gain elements from the InP substrate; and (4) placing the gain elements onto pre-defined receptor sites on the silicon PIC and bonding them using a thin adhesive or direct bonding layer. This process is heterogeneous integration in its most literal sense — it involves the integration of components from different ("hetero") material systems ("geneous") through a post-fabrication bonding step.

The WDM Elements Alone Are Monolithic — But the Die Is Not. OWS might argue that the WDM elements themselves (the ring resonator MUX/DEMUX in the LumosBridge 800G and the echelle grating MUX/DEMUX in the LumosEdge 400ZR+) are fabricated monolithically on the silicon substrate, and that the "monolithically formed" limitation should be read to apply only to the WDM element. This is essentially a restatement of OWS's proposed construction, which the Court rejected. The Court's construction unambiguously requires that the fabrication be "entirely within the same front-end-of-line and back-end-of-line semiconductor processing sequence ... without heterogeneous bonding or attachment of separately fabricated optical components." The word "entirely" and the exclusionary clause apply to the die as a whole, not merely to the WDM element in isolation. As the Court explained, the prosecution history demonstrates that the applicant intended to exclude any die that incorporates heterogeneously bonded optical components, regardless of whether the WDM element itself is monolithically fabricated.

The Presence of III-V Components Is Not a Peripheral Feature. Moreover, the III-V gain elements are not peripheral or optional components of Lumos's accused products. They are essential functional elements — the laser sources and optical amplifiers — without which the transceiver cannot operate. The micro-transfer-printed III-V gain elements are integrated onto the same silicon die that houses the WDM elements, modulators, and photodetectors. This means that the very die on which the WDM element is located also contains separately fabricated and heterogeneously bonded optical components. Under the Court's construction, such a die does not satisfy the "monolithically formed" limitation.

3. Literal Infringement Conclusion

Literal infringement of Claim 1 is not viable. Lumos's LumosBridge 800G and LumosEdge 400ZR+ products do not meet the "integrated wavelength-division multiplexing element monolithically formed on a single silicon die" limitation as construed by the Court. The micro-transfer printing of III-V gain elements onto the silicon PIC constitutes, by definition, a heterogeneous integration technique involving separately fabricated optical components bonded to the silicon substrate. This falls squarely within the exclusionary language of the Court's construction.

Because the "monolithically formed" limitation is a necessary element of Claim 1, and because Lumos's products do not satisfy this limitation, Lumos does not literally infringe Claim 1. No analysis of the remaining claim limitations is necessary, though we note for completeness that Lumos's products likely satisfy all other limitations of Claim 1 (silicon photonic integrated circuit on SOI substrate, WDM element comprising ring resonators or echelle gratings, plurality of optical modulators, plurality of photodetectors, common optical interface, and electronic control circuitry).

4. Doctrine of Equivalents Analysis

The doctrine of equivalents does not provide an alternative basis for infringement of Claim 1. Prosecution history estoppel, as established by the Supreme Court in Festo Corp. v. Fenmore & Co., 535 U.S. 722 (2002), bars OWS from asserting that Lumos's heterogeneous integration technique is equivalent to the claimed "monolithically formed" limitation.

Narrowing Amendment. During prosecution of the '678 Patent, OWS narrowed the claims by amending Claim 1 to add the "monolithically formed" limitation. This amendment was made in response to the examiner's rejection of the original claims over Chen (U.S. Patent No. 9,612,401), which disclosed a multi-channel optical transceiver with heterogeneously integrated III-V components on a silicon photonic platform. The amendment was a narrowing amendment made for purposes of patentability — i.e., to overcome the prior art rejection — and therefore triggers the presumption of prosecution history estoppel under Festo.

Express Disavowal. The applicant's prosecution remarks went beyond the mere act of amendment and expressly stated that "the present invention is distinguished from heterogeneous integration approaches such as those disclosed in Chen." This statement constitutes an unambiguous and deliberate disavowal of claim scope encompassing heterogeneous integration techniques. The applicant did not merely argue that Chen failed to anticipate the claims; the applicant defined the invention by reference to the exclusion of heterogeneous integration. This express disavowal provides an independent basis for estoppel, beyond the amendment-based estoppel under Festo. See Omega Eng'g, Inc. v. Raytek Corp., 334 F.3d 1314, 1323–26 (Fed. Cir. 2003) (disavowal of claim scope through prosecution remarks limits claim scope and forecloses doctrine of equivalents).

Presumption of Surrender. Under Festo, when a narrowing amendment is made for purposes of patentability, there is a presumption that the patentee surrendered all equivalents between the original broad claim scope and the amended narrow claim scope. Here, the original claim scope encompassed any optical transceiver "integrated on a single die," which would have included heterogeneously integrated transceivers. The amended claim scope is limited to transceivers with a WDM element "monolithically formed on a single silicon die," which excludes heterogeneous integration. The surrendered territory therefore includes, at a minimum, all heterogeneous integration techniques — including micro-transfer printing, wafer bonding, flip-chip bonding, and hybrid integration.

*Festo* Rebuttal Exceptions. Under Festo, the patentee may rebut the presumption of surrender by demonstrating that: (1) the alleged equivalent was unforeseeable at the time of the amendment; (2) the rationale underlying the narrowing amendment bore no more than a tangential relation to the equivalent in question; or (3) there is some other reason suggesting that the patentee could not reasonably have been expected to have described the insubstantial substitute in question. 535 U.S. at 740–41. None of these exceptions is available here.

Exception 1 — Unforeseeability. Micro-transfer printing of III-V gain elements onto silicon photonic integrated circuits was a well-known and actively developing technology at the time of the amendment (the relevant amendment was submitted in January 2020). By that date, numerous publications and patents described micro-transfer printing as a heterogeneous integration technique for silicon photonics, including publications by researchers at Ghent University, IMEC, and Tyndall National Institute dating to 2017 and earlier. Lumos's own technical presentations at the Optical Fiber Communication Conference ("OFC") in 2019 described micro-transfer printing as a viable manufacturing technique for co-packaged optics. The technique was clearly foreseeable at the time of the amendment.

Exception 2 — Tangential Relation. The rationale underlying the narrowing amendment bore a direct — not tangential — relation to the equivalent in question. The amendment was made specifically to distinguish heterogeneous integration techniques. Micro-transfer printing is a heterogeneous integration technique. There is no daylight between the reason for the amendment and the nature of the equivalent. The tangential relation exception is plainly inapplicable.

Exception 3 — Other Reason. There is no evidence in the record suggesting that OWS could not reasonably have been expected to draft a claim that encompassed micro-transfer printing while still overcoming the Chen prior art. OWS could have, for example, drafted claims that required the WDM element to be monolithically formed while expressly permitting heterogeneous integration of other components (such as gain elements). The applicant chose instead to draw a bright line at monolithic integration and to expressly distinguish heterogeneous approaches. This was a deliberate prosecution strategy, not an unavoidable drafting constraint. The third Festo exception does not apply.

Conclusion. The doctrine of equivalents is unavailable for the "monolithically formed" limitation. Prosecution history estoppel — arising from both the narrowing amendment and the express disavowal in the prosecution remarks — completely bars any attempt to capture Lumos's heterogeneous integration architecture as an equivalent of the claimed monolithic integration. This conclusion applies with equal force to all five asserted claims, each of which incorporates the "monolithically formed" limitation of Claim 1.

B. Claim 2 (Dependent on Claim 1)

Claim 2 of the '678 Patent reads:

2\. The multi-channel optical transceiver of claim 1, wherein the wavelength-division multiplexing element is configured to multiplex or demultiplex at least four optical channels, each optical channel having a wavelength spacing of no greater than 800 GHz.

Claim 2 depends from Claim 1 and adds a limitation specifying that the WDM element must be configured to handle at least four optical channels with a wavelength spacing of no greater than 800 GHz. Because Claim 2 incorporates all limitations of Claim 1 by dependency, it necessarily includes the "integrated wavelength-division multiplexing element monolithically formed on a single silicon die" limitation.

As established above, Lumos's accused products do not satisfy the "monolithically formed" limitation of Claim 1 under the Court's construction. Because all limitations of the independent claim must be satisfied for a dependent claim to be infringed, see Wahpeton Canvas Co. v. Frontier, Inc., 870 F.2d 1546, 1552 n.9 (Fed. Cir. 1989), the failure to satisfy the "monolithically formed" limitation is dispositive of Claim 2 infringement as well. No separate analysis of the additional Claim 2 limitation (four channels, 800 GHz spacing) is required, though we note that Lumos's products likely satisfy this additional limitation standing alone: the LumosBridge 800G employs eight channels with 400 GHz spacing, and the LumosEdge 400ZR+ employs four channels with 800 GHz spacing.

Conclusion: Infringement of Claim 2 is not viable, both literally and under the doctrine of equivalents, for the same reasons that Claim 1 infringement is not viable.

C. Claim 6 (Dependent on Claim 1)

Claim 6 of the '678 Patent reads:

6\. The multi-channel optical transceiver of claim 1, further comprising a plurality of germanium-based photodetectors monolithically integrated with the wavelength-division multiplexing element on the single silicon die and configured to detect optical signals in the C-band wavelength range.

Claim 6 depends from Claim 1 and adds a limitation requiring the transceiver to include germanium-based photodetectors monolithically integrated with the WDM element on the single silicon die, configured for C-band detection. This additional limitation further emphasizes monolithic integration, specifying that the photodetectors must also be monolithically integrated on the same die as the WDM element.

The analysis is straightforward. Claim 6 incorporates all limitations of Claim 1, including the "monolithically formed" limitation. Because Claim 1's "monolithically formed" limitation is not satisfied by Lumos's products, Claim 6 is necessarily not infringed. The additional Claim 6 limitation does not cure the deficiency in the independent claim.

We note that Lumos's products do include germanium-based photodetectors on the silicon PIC configured for C-band detection, and these photodetectors are in fact fabricated within the monolithic SOI process sequence. This means that the additional limitation of Claim 6 is likely independently met by Lumos's products. However, this is immaterial: the failure on the independent claim's critical "monolithically formed" limitation is fatal to infringement of Claim 6, regardless of whether the additional dependent claim limitation is satisfied.

Conclusion: Infringement of Claim 6 is not viable, both literally and under the doctrine of equivalents, for the same reasons that Claim 1 infringement is not viable.

D. Claim 9 (Dependent on Claim 1)

Claim 9 of the '678 Patent reads:

9\. The multi-channel optical transceiver of claim 1, wherein the multi-channel optical transceiver is configured to operate at an aggregate data rate of at least 400 gigabits per second using a coherent modulation format selected from the group consisting of dual-polarization quadrature phase-shift keying (DP-QPSK) and dual-polarization 16-state quadrature amplitude modulation (DP-16QAM).

Claim 9 depends from Claim 1 and adds a limitation specifying that the transceiver must operate at an aggregate data rate of at least 400 Gbps using a specified coherent modulation format (DP-QPSK or DP-16QAM). This limitation relates to the functional performance characteristics of the transceiver, rather than to its physical structure or fabrication process.

The analysis is identical to that for Claims 2 and 6. Claim 9 incorporates all limitations of Claim 1, including the "monolithically formed" limitation. Because Claim 1 is not infringed, Claim 9 is necessarily not infringed.

For completeness, we note that Lumos's accused products likely satisfy the additional Claim 9 limitation: the LumosBridge 800G operates at 800 Gbps (exceeding the 400 Gbps threshold) using DP-16QAM modulation, and the LumosEdge 400ZR+ operates at 400 Gbps using DP-QPSK modulation. Both products therefore meet or exceed the data rate and modulation format requirements of Claim 9. However, this does not alter the conclusion: the "monolithically formed" limitation of Claim 1 remains unsatisfied.

Conclusion: Infringement of Claim 9 is not viable, both literally and under the doctrine of equivalents, for the same reasons that Claim 1 infringement is not viable.

E. Claim 14 (Dependent on Claim 1)

Claim 14 of the '678 Patent reads:

14\. The multi-channel optical transceiver of claim 1, wherein electronic driver circuitry and transimpedance amplifier circuitry are co-integrated with the wavelength-division multiplexing element on the single silicon die, and wherein the wavelength-division multiplexing element is evanescently coupled to at least one of the plurality of optical modulators through a tapered silicon waveguide.

Claim 14 depends from Claim 1 and adds two further limitations: (1) the electronic driver circuitry and transimpedance amplifier ("TIA") circuitry must be co-integrated with the WDM element on the single silicon die; and (2) the WDM element must be evanescently coupled to at least one optical modulator through a tapered silicon waveguide. These limitations describe additional structural and functional relationships among the components on the die.

Again, the analysis follows the same pattern. Claim 14 incorporates all limitations of Claim 1, including the "monolithically formed" limitation. The failure to satisfy the "monolithically formed" limitation in Claim 1 is fatal to Claim 14, regardless of whether the additional limitations are independently met.

We note that the co-integration of electronic driver and TIA circuitry on the same silicon PIC is a feature of certain configurations of the LumosEdge 400ZR+, which includes an integrated electronic-photonic chip in some packaging variants. The LumosBridge 800G, by contrast, uses a separate electronic IC (a "driver/TIA chip") that is flip-chip bonded onto a common interposer alongside the silicon PIC, rather than co-integrated on the same die. Thus, even if the "monolithically formed" limitation were satisfied, infringement of Claim 14 would be questionable as to the LumosBridge 800G (though this issue is now moot). Regardless, the dispositive deficiency on the "monolithically formed" limitation renders this analysis academic.

Conclusion: Infringement of Claim 14 is not viable, both literally and under the doctrine of equivalents, for the same reasons that Claim 1 infringement is not viable.

V. Prosecution History Estoppel Analysis — Detailed Discussion

Because prosecution history estoppel is the critical legal doctrine that forecloses the doctrine of equivalents — OWS's only remaining theoretical path to infringement — this Section provides a detailed analysis of the prosecution history of the '678 Patent and the scope and application of estoppel.

A. The Prosecution History of the '678 Patent

The '678 Patent application (U.S. Application No. 16/015,842) was filed on June 22, 2018, and underwent a protracted prosecution spanning nearly three years before issuance on June 29, 2021. The prosecution history includes three office action rejections and three applicant responses, as well as an examiner interview summary dated October 8, 2019.

Original Claims. As originally filed, Claim 1 of the application recited "a multi-channel optical transceiver, comprising: a silicon photonic integrated circuit integrated on a single die ..." The claim did not include the phrase "monolithically formed" or any other limitation expressly requiring monolithic fabrication. The original Claim 1 was broad enough to encompass both monolithically integrated and heterogeneously integrated transceivers.

First Office Action (November 14, 2018). The examiner rejected Claims 1–18 under 35 U.S.C. § 102(a)(1) as anticipated by U.S. Patent No. 9,612,401 to Chen et al. ("Chen"). Chen discloses a multi-channel optical transceiver formed on a silicon photonic platform that includes heterogeneously integrated III-V gain elements bonded to the silicon substrate. The examiner found that Chen disclosed all limitations of the original Claim 1, including a WDM element on the silicon die, optical modulators, photodetectors, a common optical interface, and electronic control circuitry. The examiner's rejection noted that Chen's transceiver "integrates all claimed components on a single silicon die, albeit using heterogeneous bonding for the III-V gain elements."

First Response (May 14, 2019). The applicant traversed the rejection, arguing that Chen's heterogeneously integrated transceiver was distinguishable from the claimed invention. However, the applicant did not amend the claims in this response. Instead, the applicant argued that the specification's emphasis on monolithic integration and the described advantages of a "single, unified fabrication process" should be read into the claims to distinguish Chen. The applicant stated: "The claimed optical transceiver is fundamentally different from Chen's hybrid platform because it achieves integration through a monolithic fabrication process, as described throughout the specification."

Second Office Action (August 23, 2019). The examiner maintained the rejection and was unpersuaded by the applicant's arguments. The examiner stated: "While the specification describes a preference for monolithic fabrication, the claims as currently written do not require monolithic fabrication. The claim language 'integrated on a single die' is broad enough to encompass both monolithic and heterogeneous integration. The applicant has not pointed to any claim language that distinguishes the claimed invention from Chen's heterogeneously integrated transceiver." The examiner further noted that "limitations from the specification cannot be read into the claims to distinguish prior art."

Examiner Interview (October 8, 2019). An examiner interview was conducted on October 8, 2019, between the examiner (L. Rodriguez, Art Unit 2883) and the applicant's prosecuting attorney (Susan K. Whitfield of Broadmoor IP PLLC). According to the interview summary, the examiner indicated that "amending the claims to require monolithic formation of the WDM element on the silicon die could distinguish Chen, which relies on heterogeneous bonding." The applicant's representative indicated that such an amendment would be considered.

Second Response / Third Office Action Cycle (January 15, 2020). Following the examiner interview, the applicant submitted an amended set of claims on January 15, 2020. The amendment modified Claim 1 to replace "a silicon photonic integrated circuit integrated on a single die" with "a silicon photonic integrated circuit formed on a silicon-on-insulator substrate" and to add the limitation "an integrated wavelength-division multiplexing element monolithically formed on a single silicon die." The amendment also added the "wherein" clause requiring that the WDM element, modulators, photodetectors, and common optical interface be "co-integrated on the single silicon die."

Crucially, the applicant's remarks accompanying the amendment included the following statement:

> "The present invention is distinguished from heterogeneous integration approaches such as those disclosed in Chen. The claimed multi-channel optical transceiver requires a WDM element that is monolithically formed on a single silicon die, meaning fabricated within the same semiconductor processing sequence without the need for separate fabrication and bonding of optical components from different material systems. Chen's transceiver, which relies on heterogeneously bonded III-V gain elements, does not meet this limitation. Applicant respectfully submits that the amended claims are patentable over Chen and requests reconsideration and withdrawal of the rejection."

Allowance (March 30, 2021). Following this amendment and the applicant's remarks, the examiner issued a Notice of Allowance on March 30, 2021. The examiner's Reasons for Allowance stated: "The prior art of record does not teach or suggest a multi-channel optical transceiver comprising an integrated wavelength-division multiplexing element monolithically formed on a single silicon die, as recited in the amended claims. The closest prior art, Chen, discloses a heterogeneously integrated transceiver that does not meet the monolithically formed limitation."

B. Scope of Estoppel

The prosecution history establishes that the claims of the '678 Patent were narrowed by amendment for purposes of patentability. The amendment changed the claims from a broad recitation of "integrated on a single die" — which encompassed any integration technique — to "monolithically formed on a single silicon die" — which excludes non-monolithic integration techniques. This narrowing amendment was made specifically to overcome the examiner's rejection over Chen, a reference that disclosed heterogeneous integration.

Under Festo, when a patent claim is narrowed by amendment during prosecution for a reason related to patentability, there arises a presumption that the patentee has surrendered all territory between the original claim and the amended claim. 535 U.S. at 733–34. The "territory" surrendered here is substantial: it encompasses all forms of non-monolithic integration of optical components on a silicon die, including but not limited to:

• Heterogeneous integration via wafer bonding of III-V materials to silicon;

• Micro-transfer printing of separately fabricated III-V components onto silicon;

• Flip-chip bonding of III-V or other optical components onto silicon;

• Hybrid integration using edge-coupled or butt-coupled separately fabricated components;

• Any other technique involving the fabrication of optical components on a separate substrate and their subsequent attachment, bonding, or transfer to the silicon die.

Micro-transfer printing — the specific technique used by Lumos — falls squarely within this surrendered territory. It is a form of heterogeneous integration that involves the separate fabrication of III-V gain elements on an InP substrate and their subsequent transfer and bonding to the silicon PIC. The applicant's prosecution remarks expressly identified heterogeneous integration as the category of technology from which the invention was distinguished. There is no basis to argue that micro-transfer printing, as a specific form of heterogeneous integration, was excluded from the scope of the disavowal.

C. Impact on Doctrine of Equivalents

As analyzed in Section IV.A.4 above, all three Festo rebuttal exceptions are inapplicable. This Section provides additional detail supporting that conclusion.

Exception 1 — Unforeseeability. The relevant inquiry is whether the alleged equivalent (Lumos's micro-transfer-printed III-V gain elements) was "unforeseeable at the time of the amendment." Festo, 535 U.S. at 740. The amendment was submitted on January 15, 2020. At that time, micro-transfer printing of III-V photonic components onto silicon was a well-established and widely published technique. Key publications predating the amendment include:

• J. Justice et al., "Wafer-scale integration of group III-V lasers on silicon using transfer printing of epitaxial layers," Nature Photonics, vol. 6, pp. 610–614 (2012);

• J. Zhang et al., "Transfer-printing-based integration of a III-V-on-silicon distributed feedback laser," Optics Express, vol. 26, no. 7, pp. 8821–8830 (2018);

• B. Corbett et al., "Transfer print techniques for heterogeneous integration of photonic components," Progress in Quantum Electronics, vol. 52, pp. 1–17 (2017);

• Lumos Photonics, Inc., "LumosBridge Co-Packaged Optics Platform: A Micro-Transfer-Printing Approach," OFC 2019 Technical Digest, Paper M2B.4 (March 2019).

The technique was not merely foreseeable — it was known, published, and commercially deployed. The first Festo exception cannot apply.

Exception 2 — Tangential Relation. The second exception applies only where "the rationale underlying the narrowing amendment bore no more than a tangential relation to the equivalent in question." Festo, 535 U.S. at 740. Here, the rationale underlying the narrowing amendment was precisely to distinguish heterogeneous integration. The applicant's remarks stated: "The present invention is distinguished from heterogeneous integration approaches." Micro-transfer printing is a heterogeneous integration approach. The relationship between the amendment rationale and the equivalent is not tangential — it is direct and central. This exception is categorically inapplicable.

Exception 3 — Other Reason. The third exception applies where "there is some other reason suggesting that the patentee could not reasonably have been expected to have described the insubstantial substitute in question." Festo, 535 U.S. at 740–41. There is no such reason here. The applicant was aware of heterogeneous integration techniques (the amendment was made to distinguish a heterogeneous integration reference) and could have drafted claims that distinguished Chen's specific architecture without categorically excluding all heterogeneous integration. For example, the applicant could have claimed a transceiver in which "the wavelength-division multiplexing element is monolithically formed on the silicon die" without requiring that the entire die be free from heterogeneously bonded components. The applicant could also have included dependent claims that permitted certain forms of heterogeneous integration (such as gain elements) while requiring monolithic fabrication of the WDM elements. The applicant chose not to do so. This was a strategic choice, not an inability to foresee or describe the substitute.

Conclusion. Prosecution history estoppel completely bars the doctrine of equivalents for the "monolithically formed" limitation. The amendment-based estoppel under Festo and the independent disavowal-based estoppel under Omega Engineering each independently foreclose DOE arguments. OWS has no viable path to infringement of any asserted claim of the '678 Patent.

VI. Comparison with Pre-Suit Assessment

The pre-suit infringement analysis memorandum dated May 5, 2022, prepared by Christy Bauer, correctly identified the "monolithically formed on a single silicon die" limitation as a significant risk to the viability of the '678 Patent infringement theory. Ms. Bauer's memorandum devoted approximately four pages to analyzing the claim construction risk associated with this limitation and concluded that the infringement theory was viable only under OWS's broader proposed construction. The memorandum explicitly noted that "under a narrower construction — one that requires all optical components on the die to be monolithically fabricated — Lumos's micro-transfer-printed III-V architecture would likely fall outside the claim scope."

The pre-suit assessment rated the infringement probability on the '678 Patent as "moderate risk" on a scale of low, moderate, and high confidence, and flagged claim construction as "the pivotal issue" that would determine the outcome of the infringement analysis. The memorandum recommended proceeding with the '678 Patent as part of the four-patent assertion strategy, noting that the '678 Patent mapped to the core functionality of both accused products and provided significant damages leverage even if the other patents encountered difficulties.

The Markman Order has now confirmed the worst-case scenario that the pre-suit analysis anticipated. The Court's construction is, if anything, even more explicitly exclusionary than the narrow construction that Ms. Bauer hypothesized. The Court not only adopted a narrow construction of "monolithically formed," but also added express exclusionary language — "without heterogeneous bonding or attachment of separately fabricated optical components" — that leaves no room for argument.

In hindsight, the prosecution history — particularly the applicant's express remarks distinguishing heterogeneous integration — provided the Court with a compelling and largely irrefutable basis to adopt the narrow construction. The pre-suit analysis arguably underweighted the risk posed by the prosecution remarks, which were unusually specific in disavowing an entire category of technology. While the pre-suit memorandum noted the prosecution history as a risk factor, it characterized the prosecution remarks as "potentially problematic" rather than "likely dispositive." A more granular analysis of the prosecution history — specifically, the applicant's statement that "the present invention is distinguished from heterogeneous integration approaches" — would have revealed that the remarks left the Court with little discretion to adopt a broader construction. The applicant's own words effectively defined the boundary of the claims in terms that exclude Lumos's accused products.

This observation is not intended as a criticism of Ms. Bauer's analysis, which was thorough and well-reasoned within the constraints of pre-suit diligence. The risk of an adverse claim construction was clearly identified, and the recommen