PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT
CALDERWOOD & HARKNESS LLP intellectual Property Group
MEMORANDUM
TO: Alan Ngo, Responsible Partner; Omar Siddiqui, Lead Associate
FROM: Christy Bauer, Associate
DATE: February 22, 2023
RE: Post-Markman Infringement Viability Analysis — U.S. Patent No. 10,891,234 ("the '234 Patent"), Asserted Claims 1, 3, 5, 8, and 11
MATTER: Optiwave Semiconductor Corp. v. Lumos Photonics, Inc., Case No. 5:22-cv-03417-EJD (N.D. Cal.)
MATTER NUMBER: 1014-00003
PRIVILEGE DESIGNATION: This memorandum constitutes attorney work product prepared in anticipation of litigation and is protected from disclosure under Federal Rule of Civil Procedure 26(b)(3) and the work product doctrine. It also contains confidential attorney-client privileged communications. Do not distribute outside Calderwood & Harkness LLP without authorization from the Responsible Partner.
Right-click to update Table of Contents
I. Executive Summary
The purpose of this memorandum is to assess the viability of Optiwave Semiconductor Corp.'s ("OWS") infringement theories for each of the five asserted claims of U.S. Patent No. 10,891,234 ("the '234 Patent") following the Court's Markman Order entered on February 17, 2023 (Dkt. 112) (the "Markman Order"). This analysis addresses both literal infringement and infringement under the doctrine of equivalents, evaluates the downstream impact on OWS's damages framework, and provides strategic recommendations for the path forward.
On February 17, 2023, the Honorable Edward J. Torvelis issued a 34-page Markman Order construing all eleven disputed claim terms across the four patents-in-suit. The outcome was decidedly unfavorable to OWS. On eight of the eleven disputed terms, the Court adopted Defendant Lumos Photonics, Inc.'s ("Lumos") proposed construction or one substantially aligned with it. With respect to the '234 Patent, the single most critical dispute centered on the term "graded buffer layer having a continuously varying germanium concentration," which appears in independent Claim 1 and is inherited by all four asserted dependent claims (Claims 3, 5, 8, and 11).
The Court construed this term in favor of Lumos, adopting Lumos's proposed construction verbatim: "a buffer layer in which the germanium atomic fraction increases monotonically and without step changes from the silicon substrate interface to the absorber layer." OWS had proposed the materially broader construction: "a buffer layer in which the germanium concentration changes across the layer thickness." The gap between these two constructions is not one of degree but of kind, and it is dispositive of the infringement case.
Under the Court's construction, infringement of all five asserted claims (Claims 1, 3, 5, 8, and 11) is untenable against Lumos's two accused products — the LumosBridge™ 800G Optical Engine and the LumosEdge™ 400ZR+ Pluggable Transceiver. Both products utilize a step-graded SiGe buffer layer architecture with four discrete composition steps at germanium atomic fractions of 0.25, 0.50, 0.75, and 1.0. This step-graded design is expressly excluded by the "without step changes" language in the Court's construction. The step-graded structure is not a borderline case or a question of fact — Lumos's four discrete, uniform-composition epitaxial layers are, by definition, "step changes" that fall outside the scope of the claims as construed.
An analysis of the doctrine of equivalents likewise yields an unfavorable conclusion. Doctrine of equivalents arguments are foreclosed by the vitiation doctrine (application of the doctrine would render the "without step changes" language meaningless), by the specification's express disclosure of step-graded designs as a distinct embodiment separate from the claimed continuously graded embodiment, and by the strong inference that the patentee disclaimed step-graded buffers by choosing to claim only "continuously varying" when the specification disclosed both alternatives.
Recommendation: Infringement of the '234 Patent cannot be maintained in the post-Markman posture. The likelihood of prevailing on infringement at summary judgment or trial is estimated at less than five percent. This conclusion should be incorporated into the broader settlement analysis and the Board presentation scheduled for February 24, 2023.
II. Background
A. The '234 Patent Overview
The patent at issue is U.S. Patent No. 10,891,234, entitled "Germanium-on-Silicon Photodetector with Graded Buffer Layer for Enhanced Responsivity" (the "'234 Patent"). The '234 Patent was filed on September 8, 2017, and issued on January 14, 2021. The named inventors are Dr. Kevin Fong, OWS's Chief Technology Officer and Co-Founder, and Dr. Amara Osei, OWS's Principal Engineer. The patent is assigned to OWS.
The '234 Patent relates to the field of germanium-on-silicon (Ge-on-Si) photodetectors designed for use in silicon photonics platforms. A central challenge in the fabrication of Ge-on-Si photodetectors is the approximately 4.2% lattice mismatch between silicon and germanium. This mismatch, if unaddressed, results in a high density of threading dislocations in the germanium absorber layer, which degrades device performance — specifically, dark current, responsivity, and bandwidth. The '234 Patent purports to address this challenge through engineered buffer layers that manage the lattice transition between the silicon substrate and the germanium absorber layer. According to the specification, the use of a graded buffer layer in which germanium concentration varies across the layer thickness reduces threading dislocation density, thereby enhancing the responsivity of the photodetector in optical transceiver applications.
The '234 Patent contains fifteen claims in total. OWS has asserted five of those claims in this action: Claims 1, 3, 5, 8, and 11. Claim 1 is the sole independent claim among the asserted claims. Claims 3, 5, 8, and 11 are dependent claims, each depending directly from Claim 1. Each dependent claim adds one or more additional limitations but inherits all limitations of Claim 1, including the critical "graded buffer layer having a continuously varying germanium concentration" limitation. As discussed in detail below, this shared dependency on Claim 1 means that the infringement viability analysis for Claim 1 is controlling for all five asserted claims.
B. Accused Products
OWS has accused two Lumos products of infringing the '234 Patent:
LumosBridge™ 800G Optical Engine. The LumosBridge™ 800G is a co-packaged optics (CPO) engine designed for hyperscale data center switches. The product was first shipped to customers in the first quarter of 2022. Based on pricing information obtained during discovery, the unit price ranges from approximately $1,200 to $1,800, depending on configuration and volume. The LumosBridge™ 800G incorporates a germanium-on-silicon photodetector with a step-graded SiGe buffer layer as part of its receive-side optical detection circuitry. The photodetector converts incoming optical signals at 1310 nm wavelength into electrical signals for processing by the switch ASIC.
LumosEdge™ 400ZR+ Pluggable Transceiver. The LumosEdge™ 400ZR+ is a QSFP-DD form factor pluggable transceiver module designed for metro and data center interconnect (DCI) applications. The product has been in production since the third quarter of 2021 and is sold to telecommunications carriers and data center operators. Unit pricing ranges from approximately $600 to $900. The LumosEdge™ 400ZR+ incorporates the same or a substantially similar germanium-on-silicon photodetector architecture as the LumosBridge™ 800G, including a step-graded SiGe buffer layer.
Combined revenue attributable to the accused products for calendar year 2022 was approximately $148.6 million, based on financial data produced by Lumos during fact discovery (Lumos-00087234 through Lumos-00087298).
The photodetector design in both accused products utilizes a step-graded SiGe buffer comprising four discrete epitaxial layers with uniform germanium atomic fractions within each layer. The four composition steps are: (1) Si₀.₇₅Ge₀.₂₅ (Ge fraction = 0.25); (2) Si₀.₅₀Ge₀.₅₀ (Ge fraction = 0.50); (3) Si₀.₂₅Ge₀.₇₅ (Ge fraction = 0.75); and (4) Ge (Ge fraction = 1.0). Each layer has a substantially uniform composition, and the transition between layers occurs as a discrete, abrupt change in germanium concentration — that is, a step change. This architecture was confirmed through multiple sources during discovery, including Lumos's internal technical publications (Lumos-00023456 – Lumos-00023489), product datasheets publicly available on Lumos's website (Lumos-00001012 – Lumos-00001034), the deposition testimony of Lumos's Director of Photonics Engineering, Dr. Sarah Kim (Kim Dep. at 142:8–145:22), and documents produced during fact discovery reflecting Lumos's epitaxial growth recipes and fabrication process specifications (Lumos-00045678 – Lumos-00045712).
C. Procedural History Relevant to the '234 Patent Claim Construction
The claim construction proceedings in this action followed the schedule set forth in the Northern District of California Patent Local Rules and the Court's Case Management Order (Dkt. 34). The relevant procedural history as it pertains to the '234 Patent is summarized below.
On September 12, 2022, OWS served its Preliminary Claim Construction Contentions pursuant to Patent Local Rule 4-1. In those contentions, OWS proposed a broad construction of "graded buffer layer having a continuously varying germanium concentration," arguing that the term should be construed as "a buffer layer in which the germanium concentration changes across the layer thickness." This construction was designed to encompass both continuously graded and step-graded SiGe buffer architectures.
On October 3, 2022, Lumos served its Responsive Claim Construction Contentions pursuant to Patent Local Rule 4-2, proposing its narrower construction requiring the germanium atomic fraction to increase "monotonically and without step changes from the silicon substrate interface to the absorber layer."
On October 14, 2022, the parties filed their Joint Claim Construction and Prehearing Statement pursuant to Patent Local Rule 4-3 (Dkt. 67), identifying "graded buffer layer having a continuously varying germanium concentration" as one of eleven disputed claim terms across the four patents-in-suit.
Claim construction discovery closed on October 31, 2022. During claim construction discovery, Lumos noticed and conducted depositions of both named inventors of the '234 Patent — Dr. Kevin Fong and Dr. Amara Osei. Both inventors were questioned extensively regarding their understanding of the term "continuously varying" as used in the patent and in the semiconductor physics literature. Dr. Fong testified that in his view, a person of ordinary skill in the art would understand "continuously varying" to include step-graded designs as a practical implementation of a gradual composition transition (Fong Dep. at 87:4–93:18). Dr. Osei's testimony was less favorable; she acknowledged that the term "continuously varying" has a "specific technical meaning" and that "in the literature, continuously graded and step-graded are generally treated as different things," though she maintained that the patent was intended to cover both (Osei Dep. at 112:9–118:22). The Court cited Dr. Osei's testimony in the Markman Order as support for Lumos's position. (Markman Order at 18.)
On December 5, 2022, OWS filed its Opening Claim Construction Brief (Dkt. 78), a 45-page brief supported by the Declaration of Dr. Harold Whitfield of Whitfield Technical Consulting Group, Inc. Dr. Whitfield opined that a person of ordinary skill in the art of silicon photonics and semiconductor epitaxial growth would understand "continuously varying germanium concentration" to encompass both analog (truly continuous) and digital (step-graded) composition profiles, because both achieve the functional objective of managing lattice mismatch and are routinely used interchangeably in the industry.
On December 19, 2022, Lumos filed its Responsive Claim Construction Brief (Dkt. 82), also 45 pages, supported by the testimony of Dr. Raymond Ng, a technical consultant retained through Brannick Howell & Stein LLP. Dr. Ng testified that practitioners in the field of semiconductor epitaxy and silicon photonics consistently distinguish between "continuously graded" and "step-graded" SiGe buffers as fundamentally different fabrication approaches with distinct structural characteristics. Dr. Ng cited multiple peer-reviewed publications and industry standards documents demonstrating this distinction.
On January 9, 2023, OWS filed its Reply Claim Construction Brief (Dkt. 89), limited to 15 pages per the Court's standing order.
On January 19, 2023, Lumos filed a Sur-Reply brief (Dkt. 92) by leave of Court, also limited to 15 pages, addressing new arguments raised in OWS's Reply.
The Markman hearing was held on January 30, 2023, before the Honorable Edward J. Torvelis. The hearing lasted approximately six and one-half hours. Dr. Whitfield testified as OWS's technical tutorial witness, providing the Court with an overview of Ge-on-Si photodetector technology and the role of buffer layers. Dr. Ng testified as Lumos's technical tutorial witness, providing competing testimony emphasizing the well-established distinction between continuously graded and step-graded buffer designs. Both experts were subject to questioning by the Court.
On February 17, 2023, the Court issued its Markman Order (Dkt. 112), a 34-page order construing all eleven disputed claim terms. With respect to the '234 Patent's "graded buffer layer having a continuously varying germanium concentration" term, the Court adopted Lumos's proposed construction verbatim: "a buffer layer in which the germanium atomic fraction increases monotonically and without step changes from the silicon substrate interface to the absorber layer." The Court's reasoning is analyzed in detail in Section III below.
III. The Court's Claim Construction — Detailed Analysis
A. The Disputed Term and Competing Proposals
The central disputed claim term in the '234 Patent is "graded buffer layer having a continuously varying germanium concentration," which appears in the body of independent Claim 1 and is incorporated by reference into all four asserted dependent claims (Claims 3, 5, 8, and 11). This term defines the structural characteristic of the buffer layer that is the core of the '234 Patent's purported invention. The parties' competing constructions and the Court's adopted construction are set forth below.
OWS's Proposed Construction:
"A buffer layer in which the germanium concentration changes across the layer thickness."
OWS argued that this construction captures the full scope of what a person of ordinary skill in the art ("POSITA") would understand the claim language to mean. OWS's principal argument was that in the field of semiconductor epitaxial growth, the term "continuously varying" is understood to encompass both truly continuous (analog) composition gradients — where the germanium fraction changes smoothly with no abrupt transitions — and step-graded (digital) composition profiles — where the germanium fraction changes in discrete increments across multiple sublayers of uniform composition. OWS contended that both types of graded buffers achieve the same functional purpose of managing the lattice mismatch between silicon and germanium, reducing threading dislocation density, and enhancing photodetector responsivity. OWS further argued that the '234 Patent specification describes both types of buffer layers, and that limiting the claim to only the truly continuous embodiment would improperly exclude a disclosed embodiment from the claim scope. OWS relied principally on the Declaration of Dr. Whitfield in support of this position.
Lumos's Proposed Construction:
"A buffer layer in which the germanium atomic fraction increases monotonically and without step changes from the silicon substrate interface to the absorber layer."
Lumos argued that the plain and ordinary meaning of "continuously varying" requires an unbroken, smooth gradient — a composition that changes without interruption or discrete jumps. Lumos contended that the term "continuously" has a well-established meaning in both ordinary English and in the semiconductor physics literature, and that this meaning unambiguously excludes step changes. Lumos further argued that the '234 Patent specification itself distinguishes between "continuously graded" and "step-graded" buffer layers, describing them in separate paragraphs as distinct embodiments. Lumos contended that the patentee's deliberate choice to use "continuously varying" in the claims, rather than the broader term "graded" alone, reflected an intentional decision to claim only the continuous embodiment and not the step-graded alternative. Lumos relied on the testimony of Dr. Ng and on multiple academic and industry publications demonstrating the well-established distinction between continuously graded and step-graded SiGe buffers.
The Court's Adopted Construction:
The Court adopted Lumos's proposed construction verbatim:
"A buffer layer in which the germanium atomic fraction increases monotonically and without step changes from the silicon substrate interface to the absorber layer."
(Markman Order at 22.)
B. The Court's Reasoning
The Markman Order devoted approximately eight pages (pages 14–22) to the construction of this term. The Court's analysis proceeded through the standard claim construction framework established by the Federal Circuit in Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc), beginning with the claim language itself, proceeding to the specification and prosecution history, and considering extrinsic evidence only to the extent necessary to resolve ambiguity in the intrinsic record.
Plain and Ordinary Meaning. The Court began by observing that the term "continuously varying" has a plain and ordinary meaning that a person of ordinary skill in the art would readily understand. The Court stated:
"The word 'continuously' is not a term of art unique to the patent at issue. It is a word with a well-understood meaning: 'without interruption; without break.' See Webster's New World Dictionary of the American English Language (4th ed. 2001); see also Sterling Reach Industries Dictionary of Scientific and Technical Terms (7th ed. 2003) (defining 'continuous' in the context of functions and physical properties as 'having no abrupt changes or discontinuities'). When applied to a physical quantity such as germanium concentration, 'continuously varying' unambiguously requires a smooth, uninterrupted change in that quantity — not a series of discrete jumps."
(Markman Order at 15–16.)
The Court further cited the IEEE Standard Dictionary of Electrical and Electronics Terms (7th ed. 2000), which defines "continuous" in the context of signal processing and materials science as "varying without discrete steps or discontinuities." (Markman Order at 16.)
Specification Disclosure. The Court next turned to the '234 Patent specification and found that it confirmed the narrow construction. The Court identified two key passages in the specification. First, at column 4, lines 32–48, the specification describes a "continuously graded" buffer layer embodiment:
"In a preferred embodiment, the buffer layer 120 is continuously graded, such that the germanium atomic fraction increases smoothly and without interruption from approximately zero at the interface with the silicon substrate 110 to approximately one at the interface with the germanium absorber layer 130. This continuous grading may be achieved using molecular beam epitaxy (MBE) or chemical vapor deposition (CVD) techniques with dynamically adjusted precursor flow rates."
(Markman Order at 17, quoting '234 Patent, col. 4, ll. 32–48.)
Second, at column 5, lines 5–22, the specification describes a separate "step-graded" buffer layer embodiment:
"In an alternative embodiment, the buffer layer 120 may be step-graded, comprising a plurality of discrete sublayers, each sublayer having a substantially uniform germanium composition. For example, the buffer layer may comprise four sublayers with germanium atomic fractions of approximately 0.25, 0.50, 0.75, and 1.0, respectively. Step-graded buffer layers provide a simpler fabrication process but may exhibit higher threading dislocation densities at the interfaces between sublayers compared to the continuously graded embodiment."
(Markman Order at 17–18, quoting '234 Patent, col. 5, ll. 5–22.)
The Court found this treatment of the two buffer architectures to be highly significant:
"The specification's separate treatment of 'continuously graded' and 'step-graded' buffer layers — in distinct paragraphs, with distinct descriptions, and with an express comparison of their relative merits — demonstrates that the patentee understood these to be different configurations. The claims recite 'continuously varying germanium concentration,' not merely 'varying germanium concentration' or 'graded germanium concentration.' The patentee's choice of the modifier 'continuously' must be given effect. To construe 'continuously varying' to encompass step-graded designs would render the word 'continuously' superfluous and would collapse the very distinction that the specification itself draws."
(Markman Order at 18.)
Expert Testimony. The Court considered the competing expert testimony of Dr. Whitfield and Dr. Ng. The Court found Dr. Whitfield's opinion — that a POSITA would understand "continuously varying" to encompass step-graded designs as a practical subset of graded buffer technology — unpersuasive in light of the intrinsic record. The Court stated:
"Dr. Whitfield's testimony that a person of ordinary skill in the art would understand 'continuously varying' to include step-graded designs is contradicted by the specification itself, which treats these designs as distinct. The Court cannot credit extrinsic evidence that is inconsistent with the intrinsic record. See Phillips, 415 F.3d at 1318 ('While extrinsic evidence can shed useful light on the relevant art, it is less significant than the intrinsic record in determining the legally operative meaning of claim language.')."
(Markman Order at 19.)
In contrast, the Court credited Dr. Ng's testimony that practitioners in the field of silicon photonics and semiconductor epitaxy consistently distinguish between continuously graded and step-graded SiGe buffers as fundamentally different approaches with different structural characteristics, different fabrication methods, and different defect profiles. Dr. Ng cited multiple peer-reviewed publications, including articles from the Journal of Applied Physics, IEEE Photonics Technology Letters, and Semiconductor Science and Technology, all of which used the terms "continuously graded" and "step-graded" to describe mutually exclusive buffer layer designs. (Markman Order at 19–20.)
The Court also noted Dr. Osei's deposition testimony, in which the co-inventor acknowledged that "in the literature, continuously graded and step-graded are generally treated as different things." (Markman Order at 18, citing Osei Dep. at 114:3–5.) Although Dr. Osei maintained that the patent was intended to cover both, the Court found that "the subjective intent of the inventor cannot override the objective meaning of the claim language as understood through the lens of the intrinsic record." (Markman Order at 18.)
Prosecution History. The Court briefly addressed the prosecution history, noting that the '234 Patent was allowed on first office action following an examiner interview and that no narrowing amendments were made during prosecution. The Court observed that the absence of prosecution history estoppel did not compel a broader construction:
"The fact that no narrowing amendments were made during prosecution does not, standing alone, require adoption of a broad construction. The claim language and the specification provide sufficient intrinsic evidence to resolve the meaning of the disputed term. The applicant chose to use 'continuously varying' from the outset, and the specification's separate treatment of step-graded embodiments supports the conclusion that this language was deliberately selected to describe the continuous — and only the continuous — embodiment."
(Markman Order at 20–21.)
C. Impact on OWS's Infringement Theories
The practical impact of the Court's construction on OWS's infringement theories is stark and dispositive.
Under OWS's proposed construction — "a buffer layer in which the germanium concentration changes across the layer thickness" — Lumos's step-graded buffer layer, which employs four discrete composition steps transitioning from Ge fraction 0.25 to 1.0, would unambiguously satisfy the limitation. The germanium concentration indisputably "changes across the layer thickness" in Lumos's design; it simply does so in discrete steps rather than as a smooth gradient.
Under the Court's construction, however, Lumos's four discrete composition steps are textbook "step changes" — the exact structural feature that the construction expressly excludes with the phrase "without step changes." Each transition between sublayers in Lumos's buffer — from 0.25 to 0.50, from 0.50 to 0.75, and from 0.75 to 1.0 — represents an abrupt, discontinuous jump in germanium atomic fraction. The Lumos buffer is fabricated by growing four distinct epitaxial layers sequentially, each with a uniform germanium composition, using separate growth steps with different precursor ratios. There is no gradual transition between layers; the composition changes occur at the interfaces as discrete jumps.
This is not a close question. Lumos's step-graded buffer design falls squarely and definitionally outside the scope of the Court's construction. The presence of step changes is not a matter of interpretation or a question of fact requiring expert testimony — it is a structural reality confirmed by Lumos's own fabrication specifications, product datasheets, and deposition testimony.
IV. Claim-by-Claim Infringement Viability Analysis
This section analyzes the infringement viability of each of the five asserted claims of the '234 Patent against the two accused Lumos products (LumosBridge™ 800G Optical Engine and LumosEdge™ 400ZR+ Pluggable Transceiver) under the Court's claim constructions.
A. Claim 1 (Independent Claim)
1. Claim Language
Claim 1 of the '234 Patent reads as follows:
1\. A germanium-on-silicon photodetector comprising:
(a) a silicon substrate having a crystalline orientation;
(b) a germanium absorber layer disposed above the silicon substrate and configured to absorb optical radiation at a wavelength in the range of 1260 nm to 1620 nm;
(c) a graded buffer layer having a continuously varying germanium concentration, disposed between the silicon substrate and the germanium absorber layer, and configured to reduce threading dislocation density in the germanium absorber layer;
(d) a first electrical contact coupled to the germanium absorber layer; and
(e) a second electrical contact coupled to the silicon substrate, wherein the first and second electrical contacts are configured to extract a photocurrent generated in response to the absorbed optical radiation.
2. Limitation-by-Limitation Analysis
Limitation 1(a) — "a silicon substrate having a crystalline orientation." The accused Lumos products incorporate germanium-on-silicon photodetectors fabricated on silicon substrates with (100) crystalline orientation, as reflected in Lumos's fabrication process specifications (Lumos-00045678 at p. 4) and confirmed by Dr. Kim's deposition testimony (Kim Dep. at 143:15–19). This limitation is satisfied by the accused products. This element was not in dispute during claim construction proceedings.
Limitation 1(b) — "a germanium absorber layer disposed above the silicon substrate and configured to absorb optical radiation at a wavelength in the range of 1260 nm to 1620 nm." Both accused products incorporate a germanium absorber layer that is responsive to optical radiation in the O-band (1260–1360 nm) and C-band (1530–1565 nm) wavelength ranges, well within the claimed 1260 nm to 1620 nm range. Lumos's product datasheets specify operating wavelengths of 1310 nm (LumosBridge™ 800G) and 1310 nm / 1550 nm (LumosEdge™ 400ZR+). This limitation is satisfied by the accused products. This element was not in dispute during claim construction proceedings.
Limitation 1(c) — "a graded buffer layer having a continuously varying germanium concentration, disposed between the silicon substrate and the germanium absorber layer, and configured to reduce threading dislocation density in the germanium absorber layer." This is the critical limitation, and it contains the term construed by the Court.
Under the Court's construction, this limitation requires "a buffer layer in which the germanium atomic fraction increases monotonically and without step changes from the silicon substrate interface to the absorber layer." The accused Lumos products employ a step-graded SiGe buffer layer comprising four discrete sublayers with uniform germanium atomic fractions of 0.25, 0.50, 0.75, and 1.0, respectively. Each sublayer has a substantially uniform composition throughout its thickness, and the transitions between sublayers constitute abrupt, discontinuous changes in germanium concentration — i.e., "step changes." The fabrication process involves growing each sublayer sequentially using distinct precursor gas ratios in a chemical vapor deposition (CVD) chamber, as confirmed by Lumos's epitaxial growth recipe documents (Lumos-00045689 at pp. 7–12) and Dr. Kim's testimony (Kim Dep. at 147:3–149:16).
The Lumos design thus fails to satisfy this limitation under the Court's construction in two respects. First, the germanium atomic fraction does not vary "continuously" — it remains constant within each sublayer and changes only at the interfaces between sublayers. Second, and dispositively, the buffer layer exhibits step changes at each sublayer interface, which is expressly excluded by the Court's construction requiring the germanium fraction to increase "without step changes."
This limitation is not satisfied by the accused products.
Limitation 1(d) — "a first electrical contact coupled to the germanium absorber layer." The accused products incorporate metal contacts (typically aluminum or copper metallization) coupled to the germanium absorber layer to collect photocurrent. This limitation is satisfied by the accused products. This element was not in dispute.
Limitation 1(e) — "a second electrical contact coupled to the silicon substrate, wherein the first and second electrical contacts are configured to extract a photocurrent generated in response to the absorbed optical radiation." The accused products incorporate a second electrical contact coupled to the silicon substrate, completing the p-i-n or p-n junction photodetector circuit and enabling extraction of the photocurrent generated in the germanium absorber. This limitation is satisfied by the accused products. This element was not in dispute.
3. Conclusion for Claim 1
Literal infringement of Claim 1 is not viable. The accused Lumos products satisfy Limitations 1(a), 1(b), 1(d), and 1(e) of Claim 1. However, the accused products do not satisfy Limitation 1(c) — the "graded buffer layer having a continuously varying germanium concentration" limitation — under the Court's construction. Because all limitations must be met for literal infringement, Lemelson v. United States, 752 F.2d 1538, 1551 (Fed. Cir. 1985), the failure to meet Limitation 1(c) is dispositive.
Moreover, because Claim 1 is the sole independent claim from which all four asserted dependent claims (Claims 3, 5, 8, and 11) depend, the failure to establish infringement of Claim 1 is fatal to the infringement case for all five asserted claims.
B. Claim 3 (Dependent on Claim 1)
1. Claim Language
Claim 3 of the '234 Patent reads as follows:
3\. The germanium-on-silicon photodetector of claim 1, wherein the graded buffer layer has a thickness in the range of 0.5 μm to 5.0 μm and the germanium absorber layer has a responsivity of at least 0.8 A/W at a wavelength of 1310 nm.
2. Analysis. Claim 3 depends from Claim 1 and incorporates all limitations of Claim 1, including Limitation 1(c) — the "graded buffer layer having a continuously varying germanium concentration." Claim 3 adds two additional limitations: (1) that the graded buffer layer has a thickness in the range of 0.5 μm to 5.0 μm; and (2) that the germanium absorber layer has a responsivity of at least 0.8 A/W at a wavelength of 1310 nm.
With respect to the additional limitations, Lumos's step-graded buffer layer has a total thickness of approximately 2.0 μm (four sublayers of approximately 0.5 μm each), which falls within the claimed 0.5 μm to 5.0 μm range. Additionally, based on Lumos's product datasheets and the expert analysis conducted by Dr. Kessler, the responsivity of the germanium absorber in the accused products is approximately 0.9–1.0 A/W at 1310 nm, which exceeds the claimed minimum of 0.8 A/W. Thus, the additional limitations of Claim 3 appear to be satisfied by the accused products.
However, because Claim 3 inherits all limitations of Claim 1, including the "continuously varying germanium concentration" limitation, and because the accused products do not satisfy that limitation under the Court's construction (as analyzed in Section IV.A above), infringement of Claim 3 cannot be established regardless of whether the additional limitations are met.
3. Conclusion for Claim 3
Infringement of Claim 3 is not viable. The accused products fail to satisfy the inherited Limitation 1(c), which is dispositive.
C. Claim 5 (Dependent on Claim 1)
1. Claim Language
Claim 5 of the '234 Patent reads as follows:
5\. The germanium-on-silicon photodetector of claim 1, wherein the photodetector is optically coupled to a silicon waveguide structure and is configured to detect optical radiation at a wavelength in the range of 1530 nm to 1565 nm.
2. Analysis. Claim 5 depends from Claim 1 and incorporates all limitations of Claim 1. Claim 5 adds limitations specifying that (1) the photodetector is optically coupled to a silicon waveguide structure, and (2) the photodetector is configured to detect optical radiation in the C-band wavelength range of 1530 nm to 1565 nm.
With respect to the additional limitations, the LumosEdge™ 400ZR+ operates at C-band wavelengths (approximately 1550 nm) and its photodetector is integrated with silicon waveguide structures on the photonic integrated circuit (PIC), as reflected in Lumos's technical publications and product architecture documentation (Lumos-00023467 at pp. 12–15). These additional limitations appear to be satisfied by at least the LumosEdge™ product. (The LumosBridge™ 800G operates at 1310 nm, so this claim may not apply to that product in any event.)
Nevertheless, because Claim 5 inherits the "continuously varying germanium concentration" limitation from Claim 1, and that limitation is not met by the accused products under the Court's construction, infringement of Claim 5 cannot be established.
3. Conclusion for Claim 5
Infringement of Claim 5 is not viable. The accused products fail to satisfy the inherited Limitation 1(c), which is dispositive.
D. Claim 8 (Dependent on Claim 1)
1. Claim Language
Claim 8 of the '234 Patent reads as follows:
8\. The germanium-on-silicon photodetector of claim 1, further comprising an anti-reflection coating disposed on the germanium absorber layer and wherein the photodetector has a 3 dB bandwidth of at least 25 GHz.
2. Analysis. Claim 8 depends from Claim 1 and incorporates all limitations of Claim 1. Claim 8 adds limitations specifying that (1) an anti-reflection coating is disposed on the germanium absorber layer, and (2) the photodetector has a 3 dB bandwidth of at least 25 GHz.
With respect to the additional limitations, the accused Lumos products incorporate anti-reflection coatings on the germanium absorber layer to reduce surface reflections and improve quantum efficiency, as described in Lumos's fabrication specifications (Lumos-00045701 at p. 16). Both accused products are designed for high-speed data communications and their photodetectors achieve 3 dB bandwidths in excess of 25 GHz — the LumosBridge™ 800G product datasheet specifies a photodetector bandwidth of approximately 35 GHz, and the LumosEdge™ 400ZR+ specifies approximately 30 GHz. The additional limitations of Claim 8 are therefore likely satisfied.
However, as with the other dependent claims, Claim 8 inherits the fatal Limitation 1(c), and infringement cannot be established.
3. Conclusion for Claim 8
Infringement of Claim 8 is not viable. The accused products fail to satisfy the inherited Limitation 1(c), which is dispositive.
E. Claim 11 (Dependent on Claim 1)
1. Claim Language
Claim 11 of the '234 Patent reads as follows:
11\. The germanium-on-silicon photodetector of claim 1, wherein the photodetector is integrated into a multi-channel optical transceiver module comprising at least four independent receive channels, each channel configured to operate at a data rate of at least 50 Gbps per channel.
2. Analysis. Claim 11 depends from Claim 1 and incorporates all limitations of Claim 1. Claim 11 adds limitations specifying that the photodetector is integrated into a multi-channel optical transceiver module with at least four independent receive channels, each operating at a data rate of at least 50 Gbps per channel.
With respect to the additional limitations, the LumosBridge™ 800G is an 800G optical engine incorporating eight channels at 100 Gbps each (exceeding the claimed minimum of four channels at 50 Gbps each). The LumosEdge™ 400ZR+ is a 400G transceiver that, depending on configuration, incorporates four channels at 100 Gbps each. Both accused products satisfy the multi-channel and data rate limitations of Claim 11.
However, as with all other dependent claims, Claim 11 inherits the "continuously varying germanium concentration" limitation from Claim 1, and that limitation is not satisfied by the accused products under the Court's construction.
3. Conclusion for Claim 11
Infringement of Claim 11 is not viable. The accused products fail to satisfy the inherited Limitation 1(c), which is dispositive.
V. Doctrine of Equivalents Analysis
A. Framework
Even where literal infringement cannot be established, a patent holder may still prove infringement under the doctrine of equivalents. Under this doctrine, infringement may be found if the accused product contains an element that is not literally within the claim limitation but is equivalent to the claimed element. The Federal Circuit recognizes two formulations of the equivalents inquiry: (1) the function-way-result test, under which infringement exists if the accused element "performs substantially the same function, in substantially the same way, to achieve substantially the same result" as the claimed limitation, Graver Tank & Mfg. Co. v. Linde Air Products Co., 339 U.S. 605, 608 (1950); and (2) the insubstantial differences test, under which infringement exists if the differences between the accused element and the claimed limitation are insubstantial, Warner-Jenkinson Co. v. Harborline Enterprises Davis Chem. Co., 520 U.S. 17, 40 (1997).
The doctrine of equivalents is subject to several important limitations, including prosecution history estoppel, the all-limitations rule (also referred to as the vitiation doctrine), and ensnarement of the prior art. Each of these limitations is relevant here.
B. Application to the "Continuously Varying" Limitation
OWS could, in theory, argue that Lumos's step-graded SiGe buffer layer is equivalent to the claimed "graded buffer layer having a continuously varying germanium concentration" under either the function-way-result test or the insubstantial differences test.
Under the function-way-result test, the argument would proceed as follows:
• Function: The step-graded buffer performs substantially the same function as a continuously graded buffer — namely, managing the lattice mismatch between the silicon substrate and the germanium absorber layer by providing a composition transition region that accommodates the 4.2% lattice parameter difference.
• Way: The step-graded buffer operates in substantially the same way — by gradually increasing the germanium content from the substrate to the absorber, thereby distributing the lattice strain across multiple intermediate compositions rather than concentrating it at a single abrupt Si/Ge interface.
• Result: The step-graded buffer achieves substantially the same result — reducing threading dislocation density in the germanium absorber layer and thereby enhancing photodetector responsivity and reducing dark current.
While this argument has superficial appeal, it faces severe obstacles under the applicable legal doctrines, as discussed below.
C. Prosecution History Estoppel Concerns
Prosecution history estoppel bars a patentee from using the doctrine of equivalents to recapture subject matter that was surrendered during prosecution. Festo Corp. v. Fenmore & Co., 535 U.S. 722 (2002). Estoppel applies most clearly where the patentee made narrowing amendments to overcome prior art rejections, but it may also arise from arguments made during prosecution or from the intrinsic record as a whole.
In the case of the '234 Patent, the prosecution history does not present the classic estoppel scenario: the patent was allowed on first office action following an examiner interview, and no formal narrowing amendments were made. This initially appeared to insulate OWS from prosecution history estoppel — indeed, the absence of narrowing amendments was one of the factors that the pre-suit analysis identified as favorable to OWS.
However, the absence of formal prosecution history estoppel does not end the inquiry. The specification itself draws a clear and deliberate distinction between "continuously graded" and "step-graded" buffer layer designs. The patentee disclosed both embodiments in the specification but chose to claim only "continuously varying" in the claims. This constitutes a form of claim scope election: the patentee deliberately selected narrower claim language when broader language was available and when the specification disclosed the broader concept. While this may not rise to the level of formal prosecution history estoppel under Festo, it provides strong support for an argument that the patentee disclaimed step-graded buffers — an argument that Lumos would likely make at summary judgment and that the Court would likely find persuasive given the reasoning in the Markman Order.
Moreover, the Court's Markman Order itself effectively treats "continuously varying" as a term deliberately chosen to exclude step-graded designs. The Court stated that the applicant "chose to use 'continuously varying' from the outset" and that "the specification's separate treatment of step-graded embodiments supports the conclusion that this language was deliberately selected to describe the continuous — and only the continuous — embodiment." (Markman Order at 20–21.) This reasoning, while rendered in the claim construction context, provides a near-certain basis for finding that step-graded designs are outside the scope of the doctrine of equivalents as well.
D. Ensnarement / Vitiation Concerns
The most formidable obstacle to a doctrine of equivalents theory is the all-limitations rule, also known as the vitiation doctrine. Under this doctrine, the doctrine of equivalents cannot be applied in a manner that effectively eliminates or vitiates a claim limitation. Freedman Seating Co. v. Am. Seating Co., 420 F.3d 1350, 1358 (Fed. Cir. 2005). If applying the doctrine of equivalents to a particular element would render a specific claim limitation meaningless, the equivalents argument must fail. See DePuy Spine, Inc. v. Harborview Corporation Sofamor Danek, Inc., 469 F.3d 1005, 1017 (Fed. Cir. 2006).
Here, the Court's construction of the "continuously varying germanium concentration" limitation requires the germanium atomic fraction to increase "without step changes." If the doctrine of equivalents were applied to treat a step-graded buffer — which by definition contains step changes — as equivalent to a buffer that increases "without step changes," it would directly and completely vitiate the "without step changes" language in the Court's construction. The "without step changes" limitation would be rendered meaningless, as it would no longer exclude anything.
The Federal Circuit has consistently held that the doctrine of equivalents cannot be used to achieve such a result. See, e.g., Asyst Techs., Inc. v. Emtrak, Inc., 402 F.3d 1188, 1195 (Fed. Cir. 2005) ("If a theory of equivalence would entirely vitiate a particular claim element, partial or complete, the theory must fail."); Planet bingo, LLC v. VKGS LLC, 576 F. App'x 1005, 1011 (Fed. Cir. 2014) (affirming summary judgment of no infringement under doctrine of equivalents where finding equivalence would vitiate a claim limitation).
This is not a case where the doctrine of equivalents is being asked to accommodate minor, insubstantial variations at the boundary of a claim limitation. Rather, OWS would be asking the Court to find that a structure defined by its step changes is equivalent to a structure that must exist "without step changes." This is the paradigmatic case of vitiation, and it would not survive summary judgment.
E. Specification Disclosure
As noted in Section III.B above, the '234 Patent specification describes both "continuously graded" and "step-graded" buffer layer embodiments. The continuously graded embodiment is described in column 4, lines 32–48, and the step-graded embodiment is described in column 5, lines 5–22. The specification treats these as distinct configurations with different fabrication approaches and different performance characteristics (noting, for example, that step-graded buffers "may exhibit higher threading dislocation densities at the interfaces between sublayers compared to the continuously graded embodiment").
The fact that the specification discloses both designs but the claims recite only "continuously varying" supports the inference that the patentee intended to claim the continuous embodiment and not the step-graded alternative. This inference further forecloses the application of the doctrine of equivalents to encompass step-graded designs, as the patentee cannot use the doctrine to recapture subject matter that was disclosed but deliberately not claimed. See Maxwell v. J. Tarvick, Inc., 86 F.3d 1098, 1108 (Fed. Cir. 1996) (holding that where the specification distinguishes an unclaimed embodiment from the claimed embodiment, the doctrine of equivalents does not extend to the unclaimed embodiment).
F. Conclusion on Doctrine of Equivalents
For the reasons set forth above, doctrine of equivalents arguments are unavailable or extremely weak for all five asserted claims of the '234 Patent. The analysis can be summarized as follows:
1. Vitiation: Application of the doctrine of equivalents would directly vitiate the "without step changes" language in the Court's construction, rendering it meaningless. This alone is sufficient to foreclose the equivalents theory.
2. Specification disclaimer: The specification's disclosure of step-graded buffers as a distinct, unclaimed embodiment supports the inference that the patentee disclaimed step-graded designs by choosing not to claim them.
3. Prosecution history / claim scope election: Although no formal narrowing amendments were made, the patentee's deliberate choice to use "continuously varying" when the specification disclosed both approaches constitutes a form of claim scope election that further supports exclusion of step-graded designs from the equivalents analysis.
4. The Court's Markman reasoning: The Court's own rationale in the Markman Order — treating the patentee's choice of "continuously varying" as a deliberate selection to exclude step-graded designs — would almost certainly extend to bar a doctrine of equivalents theory at summary judgment.
Recommendation: Pursuing a doctrine of equivalents theory for the '234 Patent would be unlikely to survive summary judgment and would risk adverse credibility consequences with the Court if perceived as an attempt to circumvent the Markman ruling. We do not recommend advancing this theory.
VI. Impact on Damages Theory
The post-Markman infringement viability conclusion for the '234 Patent has significant implications for OWS's overall damages framework in this action.
Pre-Markman Damages Framework. Prior to the Markman Order, OWS's damages case was structured around the combined assertion of all four patents-in-suit — the '234, '567, '678, and '890 Patents. The pre-Markman damages framework was developed with input from Dr. Lawrence Kessler of Apex Claim Analytics LLC, who served as OWS's damages expert. Dr. Kessler's preliminary analysis, set forth in his draft expert report outline (work product, not yet disclosed), applied a reasonable royalty methodology under Georgia-Pacific Corp. v. U.S. Plywood Corp., 318 F. Supp. 1116 (S.D.N.Y. 1970), to calculate a hypothetical royalty rate reflecting the combined value of the four asserted patents.
The '234 Patent's Contribution to the Damages Case. The '234 Patent was considered the second-strongest patent in the portfolio from a damages perspective, behind only the '890 Patent. This assessment was based on the fact that both accused products — the LumosBridge™ 800G and the LumosEdge™ 400ZR+ — incorporate germanium-on-silicon photodetectors, meaning that the '234 Patent's claims had the broadest potential applicability across the accused product portfolio. The '567 Patent (directed to adiabatic coupler technology) was relevant primarily to the LumosEdge™ 400ZR+ product, and the '678 Patent (directed to monolithically formed modulator structures) was relevant primarily to the LumosBridge™ 800G product.
Pre-Litigation Valuation. Clearpath IP Valuation Services, Inc. (lead analyst Jennifer Armitage) conducted a pre-litigation valuation of the four asserted patents for purposes of informing OWS's Board of Directors regarding the potential value of the litigation. Clearpath valued the four asserted patents collectively at approximately $42.8 million, with the '234 Patent representing a material portion of that valuation. While the precise apportionment among the four patents is subject to the assumptions and methodology used in Clearpath's analysis, the '234 Patent's contribution was estimated at approximately 25–30% of the total portfolio value given its applicability to both accused products.
Impact of Eliminating the '234 Patent. With the '234 Patent effectively eliminated from the infringement case, the remaining damages theories must rely solely on any surviving claims under the three remaining patents-in-suit. However, as addressed in the companion memoranda being prepared concurrently with this analysis, the '567, '678, and '890 Patents face similar or more severe post-Markman viability issues. The '567 Patent was adversely affected by the Court's construction of "adiabatically coupled" (Markman Order at 23–26), and the '678 Patent was adversely affected by the Court's construction of "monolithically formed" (Markman Order at 27–30). Only the '890 Patent remains potentially viable, and its viability depends on factual issues that are still under evaluation.
Royalty Rate Impact. The pre-Markman reasonable royalty analysis applied an aggregate royalty rate of 7.5% to the $148.6 million in combined accused product revenue, yielding a preliminary damages figure of approximately $11.1 million for the '234 Patent's contribution alone (assuming 100% apportionment of the royalty to the '234 Patent, which overstates its individual contribution). The 7.5% aggregate rate was premised on the combined contribution of all four patents; eliminating the '234 Patent materially reduces the supportable royalty rate. A rate supported by only one or two surviving patents — particularly where the surviving patents may have limited applicability to only one of the two accused products — would be substantially lower.
Overall Damages Assessment. The pre-Markman damages target of exceeding $75 million (which included reasonable royalties, supplemental damages, enhanced damages for willfulness, and pre-judgment interest) is no longer supportable. The elimination of the '234 Patent, in combination with the adverse constructions affecting the '567 and '678 Patents, fundamentally undermines the damages framework. The remaining recoverable amount will depend on the viability of the '890 Patent, which is being assessed in a separate companion memorandum.
VII. Comparison with Pre-Suit Infringement Assessment
It is instructive to compare the outcome of the Markman proceedings with the pre-suit analysis that informed OWS's decision to file this action.
The pre-suit infringement analysis memorandum for the '234 Patent, dated April 25, 2022, was prepared by Omar Siddiqui and Christy Bauer and delivered to Alan Ngo and the OWS litigation team. That memorandum analyzed the infringement theories available against Lumos's accused products and identified the construction of "graded buffer layer having a continuously varying germanium concentration" as a key risk factor.
Specifically, the pre-suit memorandum identified the risk that "continuously varying germanium concentration" could be construed to exclude step-graded designs. The memorandum noted:
"The specification's disclosure of both 'continuously graded' and 'step-graded' buffer layer embodiments creates some ambiguity regarding whether the claim term 'continuously varying' is broad enough to encompass step-graded architectures. If the Court adopts a construction that limits 'continuously varying' to smooth, uninterrupted gradients — excluding discrete step changes — OWS's infringement theory against Lumos's step-graded buffer design would fail."
However, the pre-suit memorandum assessed this risk as moderate, not high. Several factors informed that assessment:
1. The memorandum noted that the specification describes both embodiments and argued that a POSITA would understand the specification's disclosure as illustrative of the types of graded buffers encompassed by the claims, not as limiting the claims to only the continuously graded embodiment.
2. The memorandum pointed to functional equivalence between step-graded and continuously graded buffers in the academic literature, noting that several peer-reviewed publications describe step-graded buffers as a practical implementation of the same grading concept.
3. The memorandum observed that the prosecution history was clean — no narrowing amendments, no examiner rejections to overcome — which was viewed as reducing the risk of a narrow construction driven by prosecution history estoppel.
4. The memorandum expressed the view that OWS's broader construction had a "reasonable probability of adoption" and estimated the likelihood at approximately 40–50%.
In hindsight, the pre-suit assessment underweighted the significance of the specification's distinct treatment of the two buffer types. The specification does not merely describe two embodiments of the same feature; it describes them in separate paragraphs, uses different terminology ("continuously graded" versus "step-graded"), and draws an explicit comparison of their relative merits. The Court relied heavily on this distinct treatment, finding it to be strong intrinsic evidence that the patentee understood these to be different configurations. The pre-suit analysis should have given greater weight to this structural feature of the specification.
Additionally, the pre-suit assessment may have underweighted the plain meaning of the word "continuously" in the context of semiconductor physics. The Court's reliance on technical dictionaries and on Dr. Ng's testimony regarding the well-established distinction in the field between "continuous" and "step" grading was foreseeable and should have been weighted more heavily in the risk assessment.
Calibration for Future Assessments. This outcome should serve as a calibration point for future pre-suit claim construction risk assessments. Where the specification describes two distinct embodiments using different terminology and the claims use language that corresponds to only one of those embodiments, the risk of a narrow construction should be assessed as high, not moderate — particularly where the claim language has a plain meaning that is consistent with the narrow construction. The functional equivalence of the two embodiments, while relevant from a technical perspective, does not overcome strong intrinsic evidence of distinct treatment in the specification.
VIII. Consideration of the '234 Patent's Prosecution History in Context
Although the prosecution history of the '234 Patent did not play a central role in the Court's claim construction analysis — the Court relied primarily on the claim language and specification — the prosecution history is nonetheless relevant to the overall assessment and warrants brief discussion, particularly in comparison to the other patents-in-suit.
Clean Prosecution History. The '234 Patent was allowed on first office action following an examiner interview conducted on March 14, 2020. No prior art rejections under 35 U.S.C. §§ 102 or 103 were issued. No rejections under 35 U.S.C. § 112 were issued. No narrowing amendments were required, and the claims as issued are substantively identical to the claims as originally filed. The examiner's reasons for allowance cited the combination of the graded buffer layer with the specific photodetector architecture and integration features recited in the claims.
This clean prosecution history initially appeared favorable to OWS, as it avoided the prosecution history estoppel issues that have plagued the other patents-in-suit. In particular:
• The '567 Patent underwent multiple rounds of rejection during prosecution. To overcome a rejection based on the Tanaka prior art reference (U.S. Patent No. 9,456,012), the applicant was required to add the limitation "adiabatically coupled" to the independent claims. This narrowing amendment created prosecution history estoppel that Lumos has invoked and that the Court's construction of "adiabatically coupled" appears to have been influenced by. (See companion memorandum on the '567 Patent, dated February 22, 2023.)
• The '678 Patent similarly required the addition of the limitation "monolithically formed" to overcome a prior art rejection. This amendment narrowed the scope of the claims and gave rise to estoppel arguments that Lumos has asserted. (See companion memorandum on the '678 Patent, dated February 22, 2023.)
The absence of analogous amendments in the '234 Patent prosecution initially appeared to leave OWS's claim scope arguments unencumbered by estoppel. However, as discussed in Section V.C above, the clean prosecution history did not prevent a narrow construction because the Court relied primarily on the plain meaning of the claim language and the specification's disclosure, rather than on prosecution history amendments.