1014-00004 · Hawthorne Vega & Cruz LLP · Validity Defense Analysis

invalidity-analysis-847.docx DOCUMENT Litigation

158 / 168 ← → ✕

Contract analysis Extract to table

CALDERWOOD & HARKNESS LLP

INTERNAL MEMORANDUM — ATTORNEY WORK PRODUCT

PRIVILEGED AND CONFIDENTIAL

November 1, 2024

TO: Case Team — Optiwave Semiconductor Corp. v. Lumicore Technologies, Inc.

FROM: Omar Siddiqui, Associate

REVIEWED BY: Wendy Schultz, Lead Associate

APPROVED BY: Alan Ngo, Partner

MATTER: Optiwave Semiconductor Corp. v. Lumicore Technologies, Inc., Case No. 2:24-cv-00319-JRG (E.D. Tex.)

SUBJECT: Analysis of Lumicore's Invalidity Contentions for U.S. Patent No. 10,234,847 ("the '847 Patent")

CONFIDENTIALITY LEGEND: ATTORNEY WORK PRODUCT — PRIVILEGED AND CONFIDENTIAL. This memorandum was prepared in anticipation of litigation and is protected by the attorney work-product doctrine. It reflects the mental impressions, conclusions, opinions, and legal theories of Calderwood & Harkness LLP attorneys. Do not distribute outside the C&H case team without prior approval from Alan Ngo. Any unauthorized dissemination, distribution, or copying of this communication is strictly prohibited. If you have received this document in error, please contact Alan Ngo immediately at (212) 555-7100.

Right-click to update Table of Contents

I. EXECUTIVE SUMMARY

The purpose of this memorandum is to analyze and assess the strengths and weaknesses of Lumicore Technologies, Inc.'s ("Lumicore's") invalidity contentions directed to U.S. Patent No. 10,234,847 (the "'847 Patent"), as set forth in Lumicore's P.R. 3-3 Invalidity Contentions served on October 14, 2024 (D.E. 36). This analysis is intended to inform case strategy, guide preparation of our validity expert's rebuttal report, and identify areas for targeted fact discovery during the remaining discovery period.

Lumicore has asserted invalidity of the '847 Patent under 35 U.S.C. §§ 102 (anticipation) and 103 (obviousness), citing a total of six prior-art references against the asserted claims — Claims 1, 3, 5, 9, and 14. Of these six references, only one (Takahashi et al.) is asserted under an anticipation theory; the remaining five references are asserted in various obviousness combinations, with Takahashi serving as the primary or anchor reference in each combination.

Overall Assessment: Lumicore's invalidity case for the '847 Patent is the weakest of its four patent challenges in this litigation. The sole anticipation reference, Takahashi et al. ("Dense WDM Silicon Photonic MUX," IEEE Photonics Technology Letters, March 2014), fails to disclose at least two claim limitations — most critically, the "low insertion loss" limitation requiring performance of less than 1.5 dB per channel. Takahashi reports insertion loss of approximately 2.8 dB per channel, nearly double the claimed threshold. This gap is not a marginal shortcoming; it represents a fundamental performance deficiency that the inventions of the '847 Patent were specifically designed to overcome.

Lumicore's obviousness combinations fare only marginally better. Each combination suffers from one or more of the following deficiencies: (i) reliance on simulated rather than measured performance data (Simonsen), (ii) cross-platform technology transplants from incompatible material systems such as indium phosphide (InP) or silicon nitride (SiN) to silicon-on-insulator (SOI) (Bergmann and Andersson), (iii) missing claim limitations requiring the assembly of three or more references (Bergmann-based combinations), and (iv) insufficient motivation-to-combine evidence that amounts to little more than hindsight reconstruction. None of the cited combinations discloses or renders obvious a silicon photonic waveguide multiplexer achieving the claimed sub-1.5 dB insertion loss performance with the specific structural features recited in the asserted claims.

We rate the likelihood of Lumicore prevailing on invalidity of the '847 Patent as low — approximately 15–20%. This assessment accounts for the inherent uncertainty of any patent litigation, the possibility that Lumicore may supplement its contentions or present favorable expert testimony, and the fact that claim construction rulings at the March 14, 2025 Markman hearing could shift the invalidity calculus in ways that are difficult to predict with certainty at this stage.

Recommendation: We recommend proceeding with preparation of Dr. Caroline Ashford's validity rebuttal report (due June 15, 2025) to address each of the six references systematically. The report should give particular emphasis to (a) the distinguishing claim limitations, especially the insertion-loss performance threshold and the specific tapered waveguide transition structure, and (b) the strong secondary considerations of non-obviousness — including commercial success of OWS's practicing products, evidence of copying by Lumicore, and failure of others in the industry to achieve comparable performance. Dr. Ashford's report should be coordinated with Gregory Falk's industry testimony on long-felt need and market conditions.

Cross-Reference to Companion Memoranda: This memorandum is one of four coordinated invalidity contentions analyses prepared for this litigation. The companion memoranda are:

• Invalidity Contentions Analysis — '193 Patent (November 3, 2024)

• Invalidity Contentions Analysis — '562 Patent (November 5, 2024)

• Invalidity Contentions Analysis — '015 Patent (November 7, 2024)

Note: The '847 Patent and the '562 Patent are in the same patent family (the '847 Patent is a continuation of the application that issued as the '562 Patent). This creates significant overlap in the prior art cited against both patents, and several of the references analyzed herein — particularly Takahashi, Bergmann, and Chen & Park — are also asserted against the '562 Patent. This memorandum should accordingly be read in conjunction with the '562 Patent invalidity analysis dated November 5, 2024, and cross-references to that companion memorandum are provided at relevant points throughout this analysis.

II. BACKGROUND ON THE '847 PATENT

A. Patent Identification and Prosecution History

The '847 Patent is titled "Multi-Channel Silicon Photonic Waveguide Multiplexer with Low Insertion Loss." It was assigned U.S. Patent No. 10,234,847 and issued on March 19, 2019. The patent is based on an application filed on September 14, 2016, and it claims the benefit of an earlier-filed parent application as described below.

The named inventors are Dr. Jian-Wei Huang (Chief Technology Officer, Optiwave Semiconductor Corp.), Dr. Nikolai Petrov (Vice President of Engineering, Photonics Division, Optiwave Semiconductor Corp.), and two additional co-inventors from OWS's photonics design team who contributed to the waveguide multiplexer development effort. The patent was originally assigned to Optiwave Semiconductor Corp. ("OWS"), which remains the current assignee. All maintenance fees are current, and the patent remains in full force and effect.

Patent prosecution was handled by Cascade IP Advisors, LLC, under the direction of Dr. Martin Chu as Lead Prosecution Attorney. During prosecution before the United States Patent and Trademark Office, the examiner issued two office actions — a non-final office action and a subsequent final office action. The key prosecution event was an amendment made in response to the non-final office action, in which OWS narrowed independent Claim 1 to require that the multiplexer exhibit a "low insertion loss" characteristic specifically defined as less than 1.5 dB per channel. This amendment was made to distinguish the claimed invention over cited prior art that disclosed waveguide multiplexers with higher insertion loss values. The examiner found this amendment, together with the specific structural limitations of the claims, sufficient to overcome the prior art rejections and allowed the application.

This prosecution history amendment is relevant to two aspects of the current litigation. First, it informs the claim construction dispute regarding the "low insertion loss" limitation (see Joint Claim Construction Statement, October 28, 2024, D.E. 40). Second, it could give rise to prosecution history estoppel under the doctrine of equivalents ("DOE") if OWS were to assert DOE for the '847 Patent claims. However, at this stage of the litigation, OWS has not asserted infringement under the doctrine of equivalents for any '847 Patent claim; all infringement allegations for the '847 Patent are under a literal infringement theory. Accordingly, prosecution history estoppel is not presently at issue for the '847 Patent, though this could change if OWS elects to add a DOE theory at a later stage.

Family Relationship: The '847 Patent is a continuation of the application that issued as U.S. Patent No. 9,678,562 (the "'562 Patent"). The '562 Patent was filed on November 22, 2014, and issued on June 13, 2017. As a continuation, the '847 Patent claims the benefit of the '562 Patent's filing date of November 22, 2014, for purposes of prior-art analysis, but only to the extent that the asserted '847 claims are adequately supported by the written description and enablement of the parent '562 specification. This priority-date issue is analyzed in detail in Section V.B of this memorandum and has significant implications for determining whether certain references qualify as prior art against the '847 Patent's claims.

Reexamination and IPR Status: As of the date of this memorandum, no request for reexamination and no petition for inter partes review ("IPR") has been filed against the '847 Patent. Lumicore has, however, reserved the right to file IPR petitions. The statutory one-year deadline for Lumicore to file an IPR petition under 35 U.S.C. § 315(b) is May 1, 2025 — one year from the date the complaint in this action was served on Lumicore. The IPR risk assessment is addressed in Section VII of this memorandum.

B. Asserted Claims

OWS has asserted five claims of the '847 Patent in this litigation: Claims 1, 3, 5, 9, and 14. These include two independent claims (Claims 1 and 14) and three dependent claims (Claims 3, 5, and 9). The asserted claims are summarized as follows:

Claim 1 (Independent — Apparatus): Claim 1 recites a multi-channel silicon photonic waveguide multiplexer comprising (a) a plurality of input waveguides configured to receive optical signals at different wavelengths, (b) an arrayed waveguide grating (AWG) structure coupled to the input waveguides, (c) a slab coupler region disposed between the input waveguides and the arrayed waveguide grating structure, and (d) an output waveguide configured to carry a multiplexed optical signal. Critically, Claim 1 further requires that the multiplexer exhibit "low insertion loss" defined as less than 1.5 dB per channel. This performance limitation is the most significant distinguishing feature of the '847 Patent over the prior art, as discussed throughout this memorandum.

Claim 3 (Dependent on Claim 1): Claim 3 adds the limitation that the waveguide multiplexer comprises at least 8 channels. This limitation specifies a minimum channel count but does not impose an upper bound, thus encompassing devices with 8, 16, 32, or more channels.

Claim 5 (Dependent on Claim 1): Claim 5 adds the limitation that the multiplexer includes a specific tapered waveguide transition region disposed between the slab coupler and the arrayed waveguide grating. The specification describes this tapered transition as a key structural innovation that reduces coupling loss at the slab-coupler-to-AWG interface, contributing to the overall low insertion loss performance of the device. This limitation is significant for invalidity purposes because no single prior-art reference cited by Lumicore discloses this specific structural feature.

Claim 9 (Dependent on Claim 1): Claim 9 recites "means for reducing crosstalk between adjacent channels." Based on our Means-Plus-Function Analysis Memorandum dated October 5, 2024, we have determined that this claim language likely invokes 35 U.S.C. § 112(f) (means-plus-function). The corresponding structure identified in the '847 Patent specification is adiabatic bends with specific geometry positioned within the arrayed waveguide grating structure. Under means-plus-function claim construction, the scope of Claim 9 is limited to the disclosed corresponding structure and its equivalents. This significantly narrows the range of prior art that can read on this limitation, because a reference must disclose either the identical adiabatic bend structure or a structural equivalent that performs substantially the same function in substantially the same way to achieve substantially the same result.

Claim 14 (Independent — Method): Claim 14 is a method claim that recites steps of multiplexing optical signals using a silicon photonic waveguide multiplexer having the structural and performance characteristics described in Claim 1. The method steps substantially track Claim 1's structural limitations in method-step form — i.e., the method includes steps of receiving optical signals through a plurality of input waveguides, coupling the signals through a slab coupler region into an arrayed waveguide grating structure, and producing a multiplexed output signal through an output waveguide, wherein the multiplexing is performed with low insertion loss of less than 1.5 dB per channel. As a method claim, Claim 14 requires that each recited step be actually performed; the mere existence of structures capable of performing the steps is insufficient. This creates an additional analytical layer for Lumicore's prior-art arguments.

Key Disputed Claim-Construction Terms Relevant to Invalidity:

The most significant claim construction dispute relevant to the invalidity analysis concerns the term "waveguide multiplexer." OWS has proposed a broad construction: "an optical device that combines multiple optical signals of different wavelengths onto a single waveguide." Lumicore has proposed a narrower construction that would limit the term to arrayed waveguide gratings incorporating a slab coupler and a phase-shifted waveguide array. The parties' respective positions are set forth in the Joint Claim Construction Statement filed October 28, 2024 (D.E. 40), and full briefing on claim construction is complete. The Markman hearing before Judge Gilstrap is scheduled for March 14, 2025.

The choice of claim construction has a significant — and strategically favorable — inverse relationship for OWS on the invalidity and infringement issues. If the Court adopts OWS's broader construction of "waveguide multiplexer," the prior art cited by Lumicore becomes even less relevant because several references teach narrower, more specific structures that do not encompass the full scope of the broader claim term, and the structural differences between the prior art and the claims become more apparent. Conversely, if the Court adopts Lumicore's narrower construction (limited to AWGs with slab coupler and phase-shifted waveguide array), the prior art may appear somewhat closer to the claims, but Lumicore's infringement case becomes substantially harder to establish because its accused products would need to meet every element of the narrower claim scope. This inverse dynamic is favorable to OWS regardless of the outcome of claim construction.

C. Accused Products Relevant to the '847 Patent

OWS has accused two Lumicore product families of infringing the '847 Patent claims:

LumiLink 400G-ZR+ Transceiver Module. This product was launched by Lumicore in the third quarter of 2022 and generated fiscal year 2023 revenue of approximately $94.8 million. The LumiLink 400G-ZR+ is a coherent optical transceiver module designed for data-center interconnect applications at distances up to 120 kilometers. OWS alleges that the 400G-ZR+ incorporates a silicon photonic waveguide multiplexer that practices each limitation of the asserted '847 Patent claims.

LumiLink 800G-DR8 Transceiver Module. This product was launched by Lumicore in the first quarter of 2023 and generated fiscal year 2023 revenue of approximately $128.4 million. The LumiLink 800G-DR8 is a parallel single-mode transceiver module designed for short-reach data-center interconnect applications. OWS alleges that the 800G-DR8 likewise incorporates a silicon photonic waveguide multiplexer that practices each limitation of the asserted '847 Patent claims.

Infringement of the '847 Patent claims by both accused products has been confirmed by independent teardown and reverse-engineering analyses conducted by Benchmark Semiconductor Analysis, LLC. Kevin Ng, Lead Engineer at Benchmark, prepared detailed teardown reports for each product: a report dated December 12, 2022 covering the LumiLink 400G-ZR+, and a report dated June 2023 covering the LumiLink 800G-DR8. Both reports identify waveguide multiplexer structures in the accused products that correspond to the structural and performance elements recited in the '847 Patent claims. These teardown reports are relevant not only to the infringement case but also to the secondary considerations of non-obviousness, as they confirm the structural similarities between Lumicore's products and the patented technology, supporting OWS's argument that Lumicore copied the patented invention.

III. OVERVIEW OF LUMICORE'S INVALIDITY CONTENTIONS

A. Summary of References Cited

Lumicore's P.R. 3-3 Invalidity Contentions were served on October 14, 2024, in compliance with the Eastern District of Texas Patent Rules. Across all four asserted patents in this litigation, Lumicore has cited a total of twenty-three prior-art references. Of those twenty-three references, six are specifically asserted against the claims of the '847 Patent. These six references form the basis of one anticipation theory and five separate obviousness combination theories.

Lumicore has retained Dr. Rajesh Sundaram as its technical expert on issues of validity and invalidity. Dr. Sundaram is a Professor of Electrical & Computer Engineering at Pacific Western University in San Diego, California. He was retained by Lumicore's litigation counsel, Hawthorne Vega & Cruz LLP. Dr. Sundaram's expert report on invalidity is due as Lumicore's rebuttal expert report on July 15, 2025, pursuant to the Court's scheduling order. We have not yet received Dr. Sundaram's report, and therefore the analysis herein is based solely on Lumicore's written invalidity contentions and the prior-art references themselves.

Lumicore's prior-art search was conducted by Marlowe Asset Management Patent Analytics, Inc., located at 1100 Peachtree Street NE, Suite 2200, Atlanta, Georgia 30309. Marlowe Asset Management is a well-known patent search firm that regularly supports defendants in technology-sector patent litigation. As discussed in Section VIII.B below, we recommend pursuing targeted discovery regarding Marlowe Asset Management's search methodology and internal communications with Lumicore's counsel, as those communications may reveal that Marlowe Asset Management identified the same deficiencies in the prior art that we have identified in this memorandum — specifically, the insertion-loss gap — but that Lumicore proceeded with its invalidity contentions notwithstanding those identified shortcomings.

B. Table of Prior-Art References Asserted Against the '847 Patent

The following table summarizes the six prior-art references asserted by Lumicore against the '847 Patent, including the type of reference, the effective date, the invalidity theory under which each is asserted, and the specific claims targeted.

All six references pre-date the '847 Patent's filing date of September 14, 2016, and therefore qualify as prior art under the applicable provisions of 35 U.S.C. § 102(a)(1) at least with respect to that filing date.

However, as noted above, the '847 Patent is a continuation of the '562 Patent (filed November 22, 2014). To the extent that certain '847 claims are entitled to the benefit of the earlier November 22, 2014 filing date based on adequate support in the parent '562 specification, references with effective dates between November 22, 2014 and September 14, 2016 would not qualify as prior art against those claims. This priority-date issue is particularly significant for two of the six references: Simonsen (filed September 17, 2015) and Andersson (published June 2015). The priority-date analysis is addressed in Section V.B of this memorandum.

IV. DETAILED ANALYSIS OF EACH PRIOR-ART REFERENCE

This section provides a detailed reference-by-reference analysis of each of the six prior-art references asserted by Lumicore against the '847 Patent. For each reference, we describe the content of the reference, summarize Lumicore's invalidity theory, and identify the strengths and weaknesses of Lumicore's arguments. We also provide preparation notes for Dr. Ashford's expert rebuttal report where appropriate.

A. Reference 1: Takahashi et al., "Dense WDM Silicon Photonic MUX" (March 2014) — Anticipation (§ 102)

1. Overview of the Reference

Takahashi et al. is a journal article published in IEEE Photonics Technology Letters in March 2014. The article describes a dense wavelength-division multiplexing (DWDM) silicon photonic multiplexer fabricated using an arrayed waveguide grating (AWG) architecture on a silicon-on-insulator (SOI) platform. The device reported in the article is a 16-channel AWG multiplexer with a channel spacing of 200 GHz, designed for potential use in telecommunications and data-center interconnect applications.

The authors report experimental characterization of the fabricated device, including measurements of insertion loss, channel crosstalk, and spectral uniformity. The reported insertion loss is approximately 2.8 dB per channel (averaged across all 16 channels), with some channel-to-channel variation. Crosstalk between adjacent channels is reported at approximately -15 dB. The article provides a description of the device architecture, including input waveguides, slab coupler regions, the arrayed waveguide grating, and output waveguides.

Critically, the Takahashi article describes a proof-of-concept laboratory device. There is no discussion of commercial implementation, thermal stability, or scalability to commercial manufacturing environments. The article does not discuss any specific techniques for reducing insertion loss below the reported 2.8 dB level, nor does it identify insertion loss as a target for further optimization. The article does not disclose any tapered waveguide transition regions between the slab coupler and the AWG, and it does not disclose any structures specifically designed for reducing crosstalk between adjacent channels (such as adiabatic bends or other mode-confining geometries).

2. Lumicore's Anticipation Theory

Lumicore contends that Takahashi anticipates Claims 1 and 14 of the '847 Patent by disclosing each element of the independent claims. Specifically, Lumicore maps the following elements:

• Takahashi's 16-channel AWG architecture is mapped to the "plurality of input waveguides" and "arrayed waveguide grating structure" limitations;

• Takahashi's described slab coupler regions (present on both the input and output sides of the AWG) are mapped to the "slab coupler region" limitation;

• Takahashi's output waveguide, which carries the multiplexed optical signal, is mapped to the "output waveguide" limitation;

• Takahashi's reported performance data is argued to satisfy the "low insertion loss" limitation.

On the final point — the insertion loss limitation — Lumicore's contention chart attempts to overcome the obvious deficiency in Takahashi's reported data by referencing a single data point at channel 8, where the measured insertion loss was approximately 1.9 dB. Lumicore appears to argue that this individual data point, or the performance at select channels, is sufficient to meet the "less than 1.5 dB per channel" limitation.

3. Analysis and Weaknesses

Critical Deficiency — Insertion Loss Limitation. Claim 1 of the '847 Patent unambiguously requires that the multiplexer exhibit "low insertion loss" defined as "less than 1.5 dB per channel." Takahashi reports insertion loss of approximately 2.8 dB per channel, averaged across the 16-channel device. This value is nearly double the claimed threshold. Even the most favorable individual channel data point cited by Lumicore — 1.9 dB at channel 8 — does not meet the less-than-1.5 dB requirement. The gap between Takahashi's reported performance and the claimed performance is substantial: a difference of approximately 1.3 dB on average, and 0.4 dB even at the single best channel.

This deficiency is fatal to Lumicore's anticipation theory. It is black-letter law that anticipation under 35 U.S.C. § 102 requires that a single prior-art reference disclose each and every limitation of the claimed invention, either expressly or inherently. Schering Corp. v. Geneva Pharms., Inc., 339 F.3d 1373, 1377 (Fed. Cir. 2003). There is no dispute that Takahashi does not disclose insertion loss of less than 1.5 dB per channel. Lumicore's attempt to satisfy this limitation by cherry-picking a single channel's performance (which itself does not meet the threshold) is legally and technically insufficient. The claim requires the recited performance characteristic for the multiplexer as a whole — not for a single isolated channel — and even if the limitation were construed on a per-channel basis, no individual channel in Takahashi's data achieves sub-1.5 dB performance.

Moreover, the insertion-loss gap is not a trivial distinction or a matter of routine optimization. In silicon photonic waveguide multiplexers, reducing insertion loss from the 2.8 dB range to below 1.5 dB represents a significant technological achievement that required the specific structural innovations disclosed in the '847 Patent — including the novel slab coupler geometry and the tapered waveguide transition between the slab coupler and the AWG. The magnitude of this performance gap, and the structural innovations needed to close it, are precisely the contributions of the '847 Patent to the art.

Tapered Waveguide Transition (Claim 5). Although Lumicore asserts Takahashi only for anticipation of Claims 1 and 14, it is worth noting that Takahashi does not disclose a tapered waveguide transition region between the slab coupler and the arrayed waveguide grating — the limitation added by dependent Claim 5. Lumicore's contention chart does not even attempt to map this limitation under the anticipation theory, confirming that Takahashi lacks this structural element. This absence further underscores the distance between Takahashi's proof-of-concept device and the '847 Patent's claimed invention.

Crosstalk Reduction Means (Claim 9). Takahashi does not disclose any structure corresponding to the "means for reducing crosstalk between adjacent channels" as recited in Claim 9. As noted above, Takahashi reports crosstalk levels of approximately -15 dB. While -15 dB crosstalk might be acceptable for certain narrowband laboratory demonstrations, it falls well below the performance levels required for the high-density, commercial-grade data-center interconnect applications contemplated by the '847 Patent, which typically demand crosstalk levels of -25 dB or better. Takahashi does not identify crosstalk as a problem to be solved, does not describe any structures designed to address crosstalk, and does not disclose adiabatic bends or any equivalent structure. Again, Lumicore does not assert Takahashi for anticipation of Claim 9, but this gap is relevant to the completeness of Takahashi as a prior-art reference.

Assessment. Takahashi does not anticipate any asserted claim of the '847 Patent. The reference fails on at least two independent limitations: the insertion loss threshold (2.8 dB vs. <1.5 dB) and the absence of a tapered waveguide transition (relevant to Claim 5 and, indirectly, to the structural completeness of the claimed multiplexer). The anticipation theory is the weakest element of Lumicore's invalidity case for the '847 Patent, and we do not anticipate that Lumicore's expert, Dr. Sundaram, will be able to rehabilitate this theory in his report. Risk level: Very Low.

Preparation Note for Dr. Ashford. In her rebuttal report, Dr. Ashford should: (i) clearly identify and quantify the insertion-loss gap between Takahashi (2.8 dB) and the <1.5 dB claim limitation; (ii) explain why this performance difference is technologically significant — i.e., it is not merely a matter of routine design choice but reflects a fundamental limitation of the Takahashi architecture that could not be overcome without the specific structural innovations claimed in the '847 Patent; (iii) demonstrate that the innovations disclosed in the '847 Patent (tapered waveguide transition, modified slab coupler geometry, adiabatic bend structures) are what enabled the sub-1.5 dB performance; and (iv) address and rebut any argument that Takahashi inherently discloses sub-1.5 dB performance or that a POSITA reading Takahashi would understand the reference as teaching such performance.

B. Reference 2: PCT/EP2015/071234 (Simonsen) — Obviousness (§ 103), Combined with Takahashi

1. Overview of the Reference

Simonsen is a PCT (Patent Cooperation Treaty) application filed on September 17, 2015, and published on March 23, 2016, under international publication rules. The applicant is a European entity, and the named inventor is Simonsen. The application describes a silicon photonic multiplexer that employs a modified slab coupler geometry intended to reduce insertion loss in arrayed waveguide grating-based multiplexer devices.

The Simonsen application reports simulated insertion loss values ranging from approximately 1.2 dB to 1.8 dB per channel, depending on channel count. Specifically, the 8-channel design achieves simulated insertion loss of approximately 1.2 dB, while the 16-channel design achieves simulated insertion loss of approximately 1.8 dB. These values are derived from computational electromagnetic simulations — specifically, finite-difference time-domain (FDTD) simulations — and are not based on measurements of a fabricated device. The Simonsen application does not report any experimental validation of the simulated performance, nor does it describe fabrication of a physical prototype.

The Simonsen application discloses both an 8-channel and a 16-channel multiplexer design. However, it does not disclose a tapered waveguide transition region between the slab coupler and the arrayed waveguide grating. It also does not disclose any crosstalk reduction structures corresponding to the "means for reducing crosstalk between adjacent channels" recited in Claim 9 of the '847 Patent, such as adiabatic bends or equivalent geometries.

2. Lumicore's Obviousness Theory

Lumicore contends that it would have been obvious to a person of ordinary skill in the art (POSITA) to combine Takahashi's AWG multiplexer architecture with Simonsen's modified slab coupler geometry to arrive at a silicon photonic multiplexer meeting the <1.5 dB insertion loss requirement of Claim 1. Lumicore asserts this combination against all five asserted claims — Claims 1, 3, 5, 9, and 14.

Lumicore's stated motivation to combine is that both references are in the same field of silicon photonic multiplexers and that a POSITA would have been motivated to incorporate Simonsen's slab coupler modification into Takahashi's AWG architecture in order to reduce insertion loss. Lumicore characterizes this as a straightforward combination of known elements to achieve a predictable result.

3. Analysis and Weaknesses

Insertion Loss — Reliance on Simulated Data. Even accepting Lumicore's proposed combination of Takahashi and Simonsen, the argument depends critically on Simonsen's reported insertion loss values — which are entirely simulated, not measured. In the field of silicon photonic device design, it is well recognized that simulated performance values are typically optimistic compared to the performance of fabricated devices. Simulation models do not fully account for fabrication variations, surface roughness, material impurities, waveguide sidewall imperfections, and other real-world effects that increase insertion loss in physical devices.

A POSITA would understand that Simonsen's simulated values of 1.2–1.8 dB would likely translate to higher measured values once the device was actually fabricated — potentially significantly higher. The gap between simulated and measured performance in silicon photonic AWG devices is well documented in the literature and can range from 0.5 dB to over 1.0 dB depending on the specific design and fabrication process. Applying this known simulation-to-fabrication gap to Simonsen's simulated values suggests that the actual fabricated performance would likely be in the range of 1.7–2.8 dB — substantially above the <1.5 dB threshold required by Claim 1.

This reliance on simulated data is a significant vulnerability for Lumicore's obviousness theory. Lumicore cannot establish that the combination of Takahashi and Simonsen would "predictably" yield the claimed performance when the performance data on which the argument is based has never been experimentally validated.

Tapered Waveguide Transition (Claim 5). Neither Takahashi nor Simonsen discloses the specific tapered waveguide transition recited in Claim 5. Lumicore's contention chart for Claim 5, when relying on the Takahashi + Simonsen combination, either leaves the tapered waveguide transition limitation unmapped or cross-references Bergmann (Reference 3) for this element. This means that for Claim 5, Lumicore actually requires a three-reference combination — Takahashi + Simonsen + Bergmann — to account for all claim limitations. The addition of a third reference introduces additional motivation-to-combine requirements and increases the hindsight reconstruction concern, as discussed below in the analysis of Bergmann.

Crosstalk Reduction Means (Claim 9). Similarly, neither Takahashi nor Simonsen discloses structures corresponding to Claim 9's "means for reducing crosstalk between adjacent channels," as construed under 35 U.S.C. § 112(f) with corresponding structure of adiabatic bends with specific geometry. Again, Lumicore relies on Bergmann for this limitation, requiring a three-reference combination for Claim 9 as well.

Motivation to Combine. Lumicore's stated motivation to combine — that "both references relate to silicon photonic multiplexers and a POSITA would combine them to reduce loss" — is a generic, conclusory assertion that is insufficient under Federal Circuit precedent. A legally adequate motivation to combine requires more than the observation that two references are in the same general field. The proponent must identify a specific teaching, suggestion, or motivation in the references, the knowledge of a POSITA, or the nature of the problem to be solved that would have led a POSITA to combine the references. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398, 418 (2007); In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006).

Lumicore's motivation-to-combine argument is further weakened by the fact that both Takahashi and Simonsen were published and publicly available for over a year before the '847 Patent's filing date, yet no one in the field — including the authors of these references and other groups working on silicon photonic multiplexers — achieved the combination and performance claimed in the '847 Patent during that period. This "failure of others" evidence cuts strongly against obviousness and supports the inference that the combination was not as straightforward as Lumicore suggests.

Assessment. The Takahashi + Simonsen combination is insufficient for Claims 5 and 9 without the addition of Bergmann (creating a three-reference combination), and the motivation to combine is weak and conclusory for all claims. The reliance on simulated data rather than measured performance is a significant vulnerability that Dr. Ashford can effectively exploit in her rebuttal report. Risk level: Low. However, this combination is marginally stronger than the anticipation theory because it at least identifies a reference (Simonsen) that addresses the insertion loss limitation in some form, even if only through simulated data. We classify the overall risk as moderate-to-low for Claim 1 and low for the remaining claims.

C. Reference 3: U.S. Patent No. 8,412,007 (Bergmann) — Obviousness (§ 103), Combined with Takahashi and/or Simonsen

1. Overview of the Reference

Bergmann is U.S. Patent No. 8,412,007, issued on April 2, 2013. The patent discloses a photonic integrated circuit comprising waveguide structures including tapered transitions and adiabatic bends. The technology context of Bergmann is critical: the patent is directed to a III-V semiconductor platform, specifically indium phosphide (InP), not to silicon photonics. InP is a fundamentally different material system from silicon-on-insulator (SOI), with different refractive index contrast, waveguide geometry, fabrication processes, and optical mode characteristics.

Bergmann describes techniques for reducing crosstalk in waveguide arrays using specific bend geometries, including adiabatic bends that gradually change the waveguide direction while maintaining optical mode confinement. The patent also discloses tapered waveguide transitions that are used to match mode sizes between different waveguide cross-sections. However, Bergmann does not disclose a waveguide multiplexer per se — the patent is focused on individual waveguide components and design techniques applicable to photonic integrated circuits generally, rather than on complete multiplexer architectures.

2. Lumicore's Obviousness Theory

Lumicore argues that Bergmann supplies the tapered waveguide transition (relevant to Claim 5) and the crosstalk reduction means (relevant to Claim 9) that are missing from Takahashi and Simonsen. Lumicore's theory is that a POSITA would have combined Bergmann's waveguide design techniques with the Takahashi/Simonsen multiplexer architecture to arrive at a device meeting the limitations of Claims 1, 5, and 9.

This requires either a two-reference combination (Takahashi + Bergmann) or, more commonly in Lumicore's contention charts, a three-reference combination (Takahashi + Simonsen + Bergmann). The three-reference combination is required when Lumicore relies on Simonsen for the insertion-loss improvement and Bergmann for the structural elements of Claims 5 and 9.

3. Analysis and Weaknesses

Different Technology Platform. The most fundamental weakness of the Bergmann combination is the platform mismatch. Bergmann is directed to III-V photonics — specifically, InP-based photonic integrated circuits. The '847 Patent is directed to silicon photonics — specifically, silicon-on-insulator (SOI) platform devices. These material systems differ in critical respects that would be well understood by a POSITA:

• Refractive index contrast: SOI waveguides have a much higher refractive index contrast (Δn ≈ 2.0 for Si/SiO₂) compared to InP waveguides (Δn ≈ 0.3 for InGaAsP/InP). This difference profoundly affects waveguide dimensions, mode confinement, bending radii, and the design of tapered transitions.

• Waveguide geometry: Due to the high index contrast, SOI waveguides are typically sub-micron in cross-section (e.g., 500 nm × 220 nm), while InP waveguides are several micrometers in dimension. Tapered transitions and adiabatic bends designed for InP waveguides cannot be directly scaled to SOI dimensions without complete redesign.

• Fabrication constraints: SOI devices are fabricated using CMOS-compatible processes (e.g., deep-UV lithography, dry etching), while InP devices use different fabrication techniques (e.g., wet etching, epitaxial regrowth). The fabrication process affects the achievable waveguide geometries and surface quality, which in turn affect insertion loss and crosstalk.

A POSITA with expertise in silicon photonics would recognize that Bergmann's InP-based design techniques cannot be straightforwardly transplanted to an SOI AWG multiplexer. The physics governing optical mode propagation, coupling, and crosstalk are fundamentally different at the dimensional and material-property scales involved. Lumicore will bear the burden of establishing that Bergmann's tapered transition and adiabatic bend techniques are "platform-agnostic" — i.e., applicable regardless of the underlying material system — and this is a contested technical point that we believe Dr. Ashford can effectively rebut.

Bergmann Does Not Disclose a Multiplexer. Bergmann's patent is directed to individual waveguide components — tapered transitions, adiabatic bends, and waveguide routing elements — not to a complete multi-channel wavelength multiplexer. There is a significant conceptual gap between knowing how to design an individual waveguide taper or bend (Bergmann's contribution) and knowing how to integrate such components into a multi-channel AWG multiplexer to achieve specific system-level performance characteristics like sub-1.5 dB insertion loss and low crosstalk (the '847 Patent's contribution). Combining Bergmann's component-level teachings with Takahashi's system-level device requires a leap that a POSITA would not make without the benefit of hindsight — knowing in advance that the specific combination of components would yield the desired system-level performance.

The Federal Circuit has cautioned against this type of hindsight reconstruction in the obviousness analysis. In re Dembiczak, 175 F.3d 994, 999 (Fed. Cir. 1999) ("It is impermissible to use the patent in suit as a blueprint for piecing together the prior art to defeat the patent."). Here, Lumicore's theory essentially requires a POSITA to identify the specific structural features of the '847 Patent (tapered slab-coupler-to-AWG transition, adiabatic bends) as solutions to the insertion loss and crosstalk problems, and then to locate those features in a component-level InP patent — a search that would only be undertaken with knowledge of the claimed invention.

Three-Reference Combination. For Claims 5 and 9, Lumicore requires Takahashi + Simonsen + Bergmann — a three-reference combination spanning two technology platforms (SOI and InP) and two levels of device hierarchy (system-level multiplexer and component-level waveguide). Courts and the Federal Circuit have recognized that as the number of references required for a combination increases, the argument for non-obviousness strengthens, particularly when the references come from different technology domains. In re Kubin, 561 F.3d 1351, 1358 (Fed. Cir. 2009); see also Wyers v. Master Lock Co., 616 F.3d 1231, 1240 (Fed. Cir. 2010). While there is no per se rule against multi-reference combinations, the burden on the party asserting obviousness to establish a credible motivation to combine increases with each additional reference, and the inference of hindsight becomes stronger.

Assessment. The Bergmann combination is the weakest of Lumicore's obviousness theories for the '847 Patent. The cross-platform transplant from InP to SOI, the component-versus-system gap, and the requirement for a three-reference combination all create significant vulnerabilities that we are confident Dr. Ashford can effectively exploit. Risk level: Very Low.

Note on Overlap with '562 Patent: Bergmann is also cited by Lumicore against the '562 Patent (see companion memo dated November 5, 2024). The analysis of Bergmann's platform mismatch set forth above applies equally to both patents. The case team should ensure consistency between Dr. Ashford's treatment of Bergmann in the '847 Patent and '562 Patent rebuttal reports.

D. Reference 4: Okamoto, "Fundamentals of Optical Waveguides" (2006 Textbook) — Obviousness (§ 103), Background Knowledge

1. Overview of the Reference

Okamoto's "Fundamentals of Optical Waveguides" is a standard textbook on optical waveguide theory, published in its second edition in 2006. The textbook is widely used in graduate-level courses in photonics and optical engineering and is a well-known reference work in the field. The textbook covers a broad range of waveguide theory topics, including mode theory, coupled-mode theory, beam propagation methods, and the design of various integrated optical devices.

Relevant to this case, the textbook includes a chapter on arrayed waveguide grating (AWG) design principles. This chapter provides a general discussion of AWG theory, including the operating principles of slab couplers, the phase relationships in waveguide arrays, insertion loss mechanisms, channel crosstalk, and spectral response characteristics. The discussion is at a theoretical and pedagogical level, covering the fundamental physics of AWG operation across multiple material platforms — including glass, polymer, and semiconductor waveguides. The textbook does not disclose a silicon photonic multiplexer specifically, nor does it provide design recipes or fabrication guidance for achieving specific performance levels on any particular platform.

Critically for purposes of the invalidity analysis: Okamoto does not disclose specific insertion loss values below 1.5 dB for any silicon photonic implementation. The textbook discusses insertion loss as a general performance parameter and identifies the physical mechanisms that contribute to insertion loss in AWG devices (e.g., mode mismatch at slab coupler interfaces, propagation loss in waveguide arrays, phase errors), but it does not teach how to achieve the specific sub-1.5 dB performance level claimed in the '847 Patent. The textbook also does not disclose tapered waveguide transitions for AWG multiplexers or crosstalk reduction structures as recited in Claim 9.

2. Lumicore's Theory

Lumicore cites Okamoto not as a primary invalidating reference but as evidence of the background knowledge available to a POSITA. The theory is that Okamoto establishes that AWG multiplexer design principles were well known in the art, thereby lowering the threshold of inventiveness required for the '847 Patent claims. Combined with Takahashi, Lumicore argues that a POSITA with knowledge of the Okamoto textbook would have known how to optimize the Takahashi design to achieve the performance levels claimed in the '847 Patent.

Lumicore asserts this combination against Claims 1, 3, and 14.

3. Analysis and Weaknesses

Textbook as Background Reference. We do not dispute that Okamoto is a well-known reference in the field, and we expect the Court to accept that a POSITA would be familiar with the general principles described in the textbook. However, familiarity with general AWG design principles does not render the specific claims of the '847 Patent obvious. The textbook discusses waveguide theory at a level of abstraction that does not teach the specific combination of structural features and performance characteristics recited in the claims. Knowing the general theory of how AWGs work is not the same as knowing how to design a specific silicon photonic AWG multiplexer that achieves sub-1.5 dB insertion loss — if it were, the sub-1.5 dB performance level would have been achieved long before OWS's invention, given that Okamoto's textbook was published in 2006.

"Mere Optimization" Argument Is Legally Vulnerable. Lumicore's reliance on Okamoto appears designed to support an argument that achieving sub-1.5 dB insertion loss was merely a matter of "routine optimization" of known design parameters. This argument is legally vulnerable under Federal Circuit precedent, which holds that the "mere optimization" theory applies only when the parameter at issue is a "result-effective variable" — i.e., a variable that a POSITA would have known to vary, and that the claimed range was achievable through routine experimentation. In re Applied Materials, Inc., 692 F.3d 1289, 1297 (Fed. Cir. 2012); In re Aller, 220 F.2d 454, 456 (CCPA 1955).

Here, the '847 Patent claims are not directed to routine optimization of a single known parameter. Achieving sub-1.5 dB insertion loss on an SOI platform required specific structural innovations — including the tapered waveguide transition between the slab coupler and the AWG, the modified slab coupler geometry, and the adiabatic bend structures — that go beyond varying a single design parameter within a known range. This is confirmed by the prosecution history, in which the examiner allowed the claims over the prior art precisely because the combination of the insertion-loss threshold and the structural limitations distinguished the claims from the prior art. The examiner's finding, while not dispositive, is entitled to a presumption of correctness under 35 U.S.C. § 282.

Assessment. Okamoto adds minimal incremental force to Lumicore's invalidity case for the '847 Patent. It establishes general background knowledge but does not supply any missing claim limitation. The textbook does not disclose the specific insertion loss values, structural features, or platform-specific design techniques claimed in the '847 Patent. Okamoto is useful to Lumicore only as a supporting reference in combination arguments, and its contribution is largely cumulative of what any POSITA would already know. Risk level: Very Low as an independent risk; marginal supporting value in combination arguments.

E. Reference 5: U.S. Patent Application Pub. No. 2015/0XXX (Andersson) — Obviousness (§ 103), Combined with Takahashi

1. Overview of the Reference

Andersson is a published U.S. patent application with a publication date of June 2015. The application discloses a multi-channel photonic device designed for telecommunications applications. Andersson describes devices with channel counts of 4, 8, and 16 channels and discusses techniques for reducing crosstalk between adjacent channels using modified waveguide spacing and mode confinement techniques.

A critical distinction is that Andersson's device is fabricated on a silicon nitride (SiN) platform, not on a silicon-on-insulator (SOI) platform. While both SiN and SOI are "silicon-compatible" platforms in the sense that they can be fabricated in CMOS-compatible foundries, they have significantly different optical properties. SiN has a lower refractive index contrast than SOI (Δn ≈ 0.5 for SiN/SiO₂ vs. Δn ≈ 2.0 for Si/SiO₂), resulting in larger waveguide dimensions, different mode profiles, and different design constraints for components such as bends, tapers, and couplers.

Andersson does not disclose insertion loss values meeting the <1.5 dB requirement of the '847 Patent claims. The application does not disclose a tapered waveguide transition region between a slab coupler and an AWG structure.

2. Lumicore's Theory

Lumicore combines Andersson with Takahashi to address Claim 3 (the 8-channel limitation) and Claim 9 (the crosstalk reduction means). Lumicore argues that Andersson's disclosed channel counts (including 8 channels) and crosstalk reduction techniques (modified waveguide spacing and mode confinement) would be combined with Takahashi's AWG architecture by a POSITA to arrive at the claimed invention.

3. Analysis and Weaknesses

Platform Mismatch. As with Bergmann, Andersson presents a platform mismatch issue — though the mismatch here is between SiN and SOI rather than between InP and SOI. While the SiN-to-SOI gap is arguably smaller than the InP-to-SOI gap (because SiN and SOI are both fabricated in CMOS-compatible processes), the two platforms still have significantly different optical properties. The refractive index contrast, waveguide dimensions, and mode confinement characteristics differ sufficiently that crosstalk reduction techniques optimized for SiN waveguides would not straightforwardly transfer to SOI waveguides without substantial redesign.

In particular, Andersson's crosstalk reduction approach relies on modified waveguide spacing — i.e., increasing the physical distance between adjacent waveguides to reduce evanescent-field coupling. This approach is feasible in the SiN platform, where waveguide dimensions are larger and device footprints are accordingly less constrained. In the SOI platform, where the high refractive index contrast enables much smaller waveguide dimensions and tighter integration, simply increasing waveguide spacing may not be a viable or desirable solution because it would increase the device footprint and negate one of the primary advantages of silicon photonics (i.e., high integration density).

Claim 3 — 8-Channel Limitation. Lumicore's argument that Andersson supplies the 8-channel limitation for Claim 3 is weak and unnecessary because Takahashi already discloses a 16-channel device. Reducing the channel count from 16 to 8 is a trivial modification that does not require a separate prior-art reference — a POSITA designing an AWG multiplexer would routinely select the channel count based on application requirements. However, this point is largely academic because Claim 3 depends on Claim 1, and therefore all of the deficiencies identified for Claim 1 carry through to Claim 3. Even if Lumicore can establish the 8-channel limitation, it cannot establish the <1.5 dB insertion loss limitation without relying on the other combinations analyzed above.

Claim 9 — Crosstalk Reduction. Andersson's crosstalk reduction technique — modified waveguide spacing — is structurally different from the "means for reducing crosstalk between adjacent channels" recited in Claim 9 of the '847 Patent. As determined in our Means-Plus-Function Analysis Memorandum (October 5, 2024), the corresponding structure for Claim 9 is adiabatic bends with specific geometry disclosed in the '847 Patent specification. Andersson's approach of modifying waveguide spacing is neither the same structure as adiabatic bends nor a structural equivalent. The two approaches reduce crosstalk through different physical mechanisms: waveguide spacing reduces crosstalk by increasing the physical distance between waveguides (and thereby reducing evanescent-field overlap), while adiabatic bends reduce crosstalk by maintaining mode confinement through gradual waveguide curvature changes that prevent mode coupling at bend points.

Under means-plus-function analysis, a prior-art reference must disclose either the identical structure or a structural equivalent that performs substantially the same function, in substantially the same way, to achieve substantially the same result. Odetics, Inc. v. Storage Tech. Corp., 185 F.3d 1259, 1267 (Fed. Cir. 1999). Andersson's waveguide spacing modification does not meet this standard.

Assessment. Andersson adds limited value to Lumicore's invalidity case. The platform mismatch and the structural differences in the crosstalk reduction approach are significant weaknesses. The 8-channel limitation is easily met by other means and does not independently strengthen Lumicore's position. Risk level: Low.

F. Reference 6: Chen & Park, "Low-Loss AWG Design on SOI" (ECOC 2014 Conference Paper) — Obviousness (§ 103), Combined with Takahashi

1. Overview of the Reference

Chen & Park is a conference paper presented at the European Conference on Optical Communication (ECOC) in September 2014. The paper describes an arrayed waveguide grating (AWG) multiplexer design fabricated on a silicon-on-insulator (SOI) platform with a focus on reducing insertion loss. The authors report measured insertion loss of approximately 1.8–2.2 dB per channel, representing an improvement over earlier SOI AWG demonstrations but still above the <1.5 dB threshold claimed in the '847 Patent.

Chen & Park discuss the use of optimized slab coupler geometry and waveguide width tapering as techniques to reduce insertion loss. However, the waveguide width tapering described in the paper occurs at the input/output waveguide interfaces — i.e., at the points where the input and output waveguides connect to the slab coupler — and serves the purpose of mode-size matching to improve coupling efficiency at those interfaces. This is distinct from the "tapered waveguide transition region between the slab coupler and the arrayed waveguide grating" recited in Claim 5 of the '847 Patent, which is located at a different position in the device (at the interface between the slab coupler and the arrayed waveguide grating, not at the input/output waveguide interfaces) and serves a different purpose (reducing coupling loss at the slab-coupler-to-AWG interface, not input/output mode matching).

The Chen & Park paper does not explicitly disclose insertion loss below 1.5 dB per channel. The best reported performance is approximately 1.8 dB for the lowest-loss channels, with most channels exhibiting loss in the 2.0–2.2 dB range.

2. Lumicore's Theory

Lumicore combines Chen & Park with Takahashi, arguing that application of Chen & Park's loss-reduction techniques (optimized slab coupler geometry and waveguide width tapering) to Takahashi's AWG design would render Claims 1, 5, and 14 obvious. Lumicore argues that the combination of loss-reduction techniques from both references would predictably yield sub-1.5 dB insertion loss.

Lumicore's motivation-to-combine argument is that both references address the same problem — reducing insertion loss in SOI-based AWG multiplexers — and that a POSITA would naturally combine the best techniques from both references.

3. Analysis and Weaknesses

Insertion Loss Gap Persists. Chen & Park report measured insertion loss of 1.8–2.2 dB per channel — an improvement over Takahashi's 2.8 dB but still above the <1.5 dB threshold required by Claim 1. Combining Takahashi (2.8 dB) with Chen & Park's loss-reduction techniques might plausibly bring the insertion loss value closer to 1.8 dB but does not establish that a POSITA would have achieved sub-1.5 dB performance. The gap between Chen & Park's best results (1.8 dB) and the <1.5 dB claim limitation — a difference of at least 0.3 dB — is significant in the context of silicon photonic waveguide design, where each incremental 0.1 dB reduction in insertion loss becomes progressively more difficult to achieve. The "last mile" from 1.8 dB to below 1.5 dB required structural innovations beyond what Chen & Park or Takahashi disclosed.

There is no evidence in either reference that the sub-1.5 dB threshold was within the reach of routine optimization. Neither Chen & Park nor Takahashi identifies the <1.5 dB level as a design target or suggests that further incremental improvements using their disclosed techniques would reach that level. To the contrary, Chen & Park's paper discusses the challenges of further reducing insertion loss in SOI AWG devices and identifies several loss mechanisms that are difficult to mitigate with the techniques available at the time.

Tapered Waveguide Transition (Claim 5). Chen & Park disclose "waveguide width tapering" in a general sense, but the tapering described in their paper is structurally and functionally distinct from the tapered waveguide transition recited in Claim 5 of the '847 Patent. Specifically:

• Location: Chen & Park's tapering occurs at the input/output waveguide interfaces (where the single-mode waveguides connect to the slab coupler). The '847 Patent's tapered transition is located between the slab coupler and the arrayed waveguide grating (on the opposite side of the slab coupler from the input/output waveguides).

• Purpose: Chen & Park's tapering is designed for mode-size matching — i.e., gradually expanding the waveguide width to match the slab coupler's mode field profile. The '847 Patent's tapered transition is designed to reduce coupling loss at the slab-coupler-to-AWG interface by providing a gradual transition in waveguide geometry that minimizes reflections and scattering.

• Design parameters: Because the two tapers are at different locations and serve different functions, their design parameters (taper length, width ratio, profile shape) are necessarily different. A POSITA would not conflate the two.

Lumicore's mapping of Chen & Park's waveguide tapering to the Claim 5 limitatio

CALDERWOOD & HARKNESS LLP
INTERNAL MEMORANDUM — ATTORNEY WORK PRODUCT
PRIVILEGED AND CONFIDENTIAL
November 1, 2024
TO: Case Team — Optiwave Semiconductor Corp. v. Lumicore Technologies, Inc.
FROM: Omar Siddiqui, Associate
REVIEWED BY: Wendy Schultz, Lead Associate
APPROVED BY: Alan Ngo, Partner
MATTER: Optiwave Semiconductor Corp. v. Lumicore Technologies, Inc., Case No. 2:24-cv-00319-JRG (E.D. Tex.)
SUBJECT: Analysis of Lumicore's Invalidity Contentions for U.S. Patent No. 10,234,847 ("the '847 Patent")
CONFIDENTIALITY LEGEND: ATTORNEY WORK PRODUCT — PRIVILEGED AND CONFIDENTIAL. This memorandum was prepared in anticipation of litigation and is protected by the attorney work-product doctrine. It reflects the mental impressions, conclusions, opinions, and legal theories of Calderwood & Harkness LLP attorneys. Do not distribute outside the C&H case team without prior approval from Alan Ngo. Any unauthorized dissemination, distribution, or copying of this communication is strictly prohibited. If you have received this document in error, please contact Alan Ngo immediately at (212) 555-7100.
Right-click to update Table of Contents
I. EXECUTIVE SUMMARY
The purpose of this memorandum is to analyze and assess the strengths and weaknesses of Lumicore Technologies, Inc.'s ("Lumicore's") invalidity contentions directed to U.S. Patent No. 10,234,847 (the "'847 Patent"), as set forth in Lumicore's P.R. 3-3 Invalidity Contentions served on October 14, 2024 (D.E. 36). This analysis is intended to inform case strategy, guide preparation of our validity expert's rebuttal report, and identify areas for targeted fact discovery during the remaining discovery period.
Lumicore has asserted invalidity of the '847 Patent under 35 U.S.C. §§ 102 (anticipation) and 103 (obviousness), citing a total of six prior-art references against the asserted claims — Claims 1, 3, 5, 9, and 14. Of these six references, only one (Takahashi et al.) is asserted under an anticipation theory; the remaining five references are asserted in various obviousness combinations, with Takahashi serving as the primary or anchor reference in each combination.
Overall Assessment: Lumicore's invalidity case for the '847 Patent is the weakest of its four patent challenges in this litigation. The sole anticipation reference, Takahashi et al. ("Dense WDM Silicon Photonic MUX," IEEE Photonics Technology Letters, March 2014), fails to disclose at least two claim limitations — most critically, the "low insertion loss" limitation requiring performance of less than 1.5 dB per channel. Takahashi reports insertion loss of approximately 2.8 dB per channel, nearly double the claimed threshold. This gap is not a marginal shortcoming; it represents a fundamental performance deficiency that the inventions of the '847 Patent were specifically designed to overcome.
Lumicore's obviousness combinations fare only marginally better. Each combination suffers from one or more of the following deficiencies: (i) reliance on simulated rather than measured performance data (Simonsen), (ii) cross-platform technology transplants from incompatible material systems such as indium phosphide (InP) or silicon nitride (SiN) to silicon-on-insulator (SOI) (Bergmann and Andersson), (iii) missing claim limitations requiring the assembly of three or more references (Bergmann-based combinations), and (iv) insufficient motivation-to-combine evidence that amounts to little more than hindsight reconstruction. None of the cited combinations discloses or renders obvious a silicon photonic waveguide multiplexer achieving the claimed sub-1.5 dB insertion loss performance with the specific structural features recited in the asserted claims.
We rate the likelihood of Lumicore prevailing on invalidity of the '847 Patent as low — approximately 15–20%. This assessment accounts for the inherent uncertainty of any patent litigation, the possibility that Lumicore may supplement its contentions or present favorable expert testimony, and the fact that claim construction rulings at the March 14, 2025 Markman hearing could shift the invalidity calculus in ways that are difficult to predict with certainty at this stage.
Recommendation: We recommend proceeding with preparation of Dr. Caroline Ashford's validity rebuttal report (due June 15, 2025) to address each of the six references systematically. The report should give particular emphasis to (a) the distinguishing claim limitations, especially the insertion-loss performance threshold and the specific tapered waveguide transition structure, and (b) the strong secondary considerations of non-obviousness — including commercial success of OWS's practicing products, evidence of copying by Lumicore, and failure of others in the industry to achieve comparable performance. Dr. Ashford's report should be coordinated with Gregory Falk's industry testimony on long-felt need and market conditions.
Cross-Reference to Companion Memoranda: This memorandum is one of four coordinated invalidity contentions analyses prepared for this litigation. The companion memoranda are:
• Invalidity Contentions Analysis — '193 Patent (November 3, 2024)
• Invalidity Contentions Analysis — '562 Patent (November 5, 2024)
• Invalidity Contentions Analysis — '015 Patent (November 7, 2024)
Note: The '847 Patent and the '562 Patent are in the same patent family (the '847 Patent is a continuation of the application that issued as the '562 Patent). This creates significant overlap in the prior art cited against both patents, and several of the references analyzed herein — particularly Takahashi, Bergmann, and Chen & Park — are also asserted against the '562 Patent. This memorandum should accordingly be read in conjunction with the '562 Patent invalidity analysis dated November 5, 2024, and cross-references to that companion memorandum are provided at relevant points throughout this analysis.
II. BACKGROUND ON THE '847 PATENT
A. Patent Identification and Prosecution History
The '847 Patent is titled "Multi-Channel Silicon Photonic Waveguide Multiplexer with Low Insertion Loss." It was assigned U.S. Patent No. 10,234,847 and issued on March 19, 2019. The patent is based on an application filed on September 14, 2016, and it claims the benefit of an earlier-filed parent application as described below.
The named inventors are Dr. Jian-Wei Huang (Chief Technology Officer, Optiwave Semiconductor Corp.), Dr. Nikolai Petrov (Vice President of Engineering, Photonics Division, Optiwave Semiconductor Corp.), and two additional co-inventors from OWS's photonics design team who contributed to the waveguide multiplexer development effort. The patent was originally assigned to Optiwave Semiconductor Corp. ("OWS"), which remains the current assignee. All maintenance fees are current, and the patent remains in full force and effect.
Patent prosecution was handled by Cascade IP Advisors, LLC, under the direction of Dr. Martin Chu as Lead Prosecution Attorney. During prosecution before the United States Patent and Trademark Office, the examiner issued two office actions — a non-final office action and a subsequent final office action. The key prosecution event was an amendment made in response to the non-final office action, in which OWS narrowed independent Claim 1 to require that the multiplexer exhibit a "low insertion loss" characteristic specifically defined as less than 1.5 dB per channel. This amendment was made to distinguish the claimed invention over cited prior art that disclosed waveguide multiplexers with higher insertion loss values. The examiner found this amendment, together with the specific structural limitations of the claims, sufficient to overcome the prior art rejections and allowed the application.
This prosecution history amendment is relevant to two aspects of the current litigation. First, it informs the claim construction dispute regarding the "low insertion loss" limitation (see Joint Claim Construction Statement, October 28, 2024, D.E. 40). Second, it could give rise to prosecution history estoppel under the doctrine of equivalents ("DOE") if OWS were to assert DOE for the '847 Patent claims. However, at this stage of the litigation, OWS has not asserted infringement under the doctrine of equivalents for any '847 Patent claim; all infringement allegations for the '847 Patent are under a literal infringement theory. Accordingly, prosecution history estoppel is not presently at issue for the '847 Patent, though this could change if OWS elects to add a DOE theory at a later stage.
Family Relationship: The '847 Patent is a continuation of the application that issued as U.S. Patent No. 9,678,562 (the "'562 Patent"). The '562 Patent was filed on November 22, 2014, and issued on June 13, 2017. As a continuation, the '847 Patent claims the benefit of the '562 Patent's filing date of November 22, 2014, for purposes of prior-art analysis, but only to the extent that the asserted '847 claims are adequately supported by the written description and enablement of the parent '562 specification. This priority-date issue is analyzed in detail in Section V.B of this memorandum and has significant implications for determining whether certain references qualify as prior art against the '847 Patent's claims.
Reexamination and IPR Status: As of the date of this memorandum, no request for reexamination and no petition for inter partes review ("IPR") has been filed against the '847 Patent. Lumicore has, however, reserved the right to file IPR petitions. The statutory one-year deadline for Lumicore to file an IPR petition under 35 U.S.C. § 315(b) is May 1, 2025 — one year from the date the complaint in this action was served on Lumicore. The IPR risk assessment is addressed in Section VII of this memorandum.
B. Asserted Claims
OWS has asserted five claims of the '847 Patent in this litigation: Claims 1, 3, 5, 9, and 14. These include two independent claims (Claims 1 and 14) and three dependent claims (Claims 3, 5, and 9). The asserted claims are summarized as follows:
Claim 1 (Independent — Apparatus): Claim 1 recites a multi-channel silicon photonic waveguide multiplexer comprising (a) a plurality of input waveguides configured to receive optical signals at different wavelengths, (b) an arrayed waveguide grating (AWG) structure coupled to the input waveguides, (c) a slab coupler region disposed between the input waveguides and the arrayed waveguide grating structure, and (d) an output waveguide configured to carry a multiplexed optical signal. Critically, Claim 1 further requires that the multiplexer exhibit "low insertion loss" defined as less than 1.5 dB per channel. This performance limitation is the most significant distinguishing feature of the '847 Patent over the prior art, as discussed throughout this memorandum.
Claim 3 (Dependent on Claim 1): Claim 3 adds the limitation that the waveguide multiplexer comprises at least 8 channels. This limitation specifies a minimum channel count but does not impose an upper bound, thus encompassing devices with 8, 16, 32, or more channels.
Claim 5 (Dependent on Claim 1): Claim 5 adds the limitation that the multiplexer includes a specific tapered waveguide transition region disposed between the slab coupler and the arrayed waveguide grating. The specification describes this tapered transition as a key structural innovation that reduces coupling loss at the slab-coupler-to-AWG interface, contributing to the overall low insertion loss performance of the device. This limitation is significant for invalidity purposes because no single prior-art reference cited by Lumicore discloses this specific structural feature.
Claim 9 (Dependent on Claim 1): Claim 9 recites "means for reducing crosstalk between adjacent channels." Based on our Means-Plus-Function Analysis Memorandum dated October 5, 2024, we have determined that this claim language likely invokes 35 U.S.C. § 112(f) (means-plus-function). The corresponding structure identified in the '847 Patent specification is adiabatic bends with specific geometry positioned within the arrayed waveguide grating structure. Under means-plus-function claim construction, the scope of Claim 9 is limited to the disclosed corresponding structure and its equivalents. This significantly narrows the range of prior art that can read on this limitation, because a reference must disclose either the identical adiabatic bend structure or a structural equivalent that performs substantially the same function in substantially the same way to achieve substantially the same result.
Claim 14 (Independent — Method): Claim 14 is a method claim that recites steps of multiplexing optical signals using a silicon photonic waveguide multiplexer having the structural and performance characteristics described in Claim 1. The method steps substantially track Claim 1's structural limitations in method-step form — i.e., the method includes steps of receiving optical signals through a plurality of input waveguides, coupling the signals through a slab coupler region into an arrayed waveguide grating structure, and producing a multiplexed output signal through an output waveguide, wherein the multiplexing is performed with low insertion loss of less than 1.5 dB per channel. As a method claim, Claim 14 requires that each recited step be actually performed; the mere existence of structures capable of performing the steps is insufficient. This creates an additional analytical layer for Lumicore's prior-art arguments.
Key Disputed Claim-Construction Terms Relevant to Invalidity:
The most significant claim construction dispute relevant to the invalidity analysis concerns the term "waveguide multiplexer." OWS has proposed a broad construction: "an optical device that combines multiple optical signals of different wavelengths onto a single waveguide." Lumicore has proposed a narrower construction that would limit the term to arrayed waveguide gratings incorporating a slab coupler and a phase-shifted waveguide array. The parties' respective positions are set forth in the Joint Claim Construction Statement filed October 28, 2024 (D.E. 40), and full briefing on claim construction is complete. The Markman hearing before Judge Gilstrap is scheduled for March 14, 2025.
The choice of claim construction has a significant — and strategically favorable — inverse relationship for OWS on the invalidity and infringement issues. If the Court adopts OWS's broader construction of "waveguide multiplexer," the prior art cited by Lumicore becomes even less relevant because several references teach narrower, more specific structures that do not encompass the full scope of the broader claim term, and the structural differences between the prior art and the claims become more apparent. Conversely, if the Court adopts Lumicore's narrower construction (limited to AWGs with slab coupler and phase-shifted waveguide array), the prior art may appear somewhat closer to the claims, but Lumicore's infringement case becomes substantially harder to establish because its accused products would need to meet every element of the narrower claim scope. This inverse dynamic is favorable to OWS regardless of the outcome of claim construction.
C. Accused Products Relevant to the '847 Patent
OWS has accused two Lumicore product families of infringing the '847 Patent claims:
LumiLink 400G-ZR+ Transceiver Module. This product was launched by Lumicore in the third quarter of 2022 and generated fiscal year 2023 revenue of approximately $94.8 million. The LumiLink 400G-ZR+ is a coherent optical transceiver module designed for data-center interconnect applications at distances up to 120 kilometers. OWS alleges that the 400G-ZR+ incorporates a silicon photonic waveguide multiplexer that practices each limitation of the asserted '847 Patent claims.
LumiLink 800G-DR8 Transceiver Module. This product was launched by Lumicore in the first quarter of 2023 and generated fiscal year 2023 revenue of approximately $128.4 million. The LumiLink 800G-DR8 is a parallel single-mode transceiver module designed for short-reach data-center interconnect applications. OWS alleges that the 800G-DR8 likewise incorporates a silicon photonic waveguide multiplexer that practices each limitation of the asserted '847 Patent claims.
Infringement of the '847 Patent claims by both accused products has been confirmed by independent teardown and reverse-engineering analyses conducted by Benchmark Semiconductor Analysis, LLC. Kevin Ng, Lead Engineer at Benchmark, prepared detailed teardown reports for each product: a report dated December 12, 2022 covering the LumiLink 400G-ZR+, and a report dated June 2023 covering the LumiLink 800G-DR8. Both reports identify waveguide multiplexer structures in the accused products that correspond to the structural and performance elements recited in the '847 Patent claims. These teardown reports are relevant not only to the infringement case but also to the secondary considerations of non-obviousness, as they confirm the structural similarities between Lumicore's products and the patented technology, supporting OWS's argument that Lumicore copied the patented invention.
III. OVERVIEW OF LUMICORE'S INVALIDITY CONTENTIONS
A. Summary of References Cited
Lumicore's P.R. 3-3 Invalidity Contentions were served on October 14, 2024, in compliance with the Eastern District of Texas Patent Rules. Across all four asserted patents in this litigation, Lumicore has cited a total of twenty-three prior-art references. Of those twenty-three references, six are specifically asserted against the claims of the '847 Patent. These six references form the basis of one anticipation theory and five separate obviousness combination theories.
Lumicore has retained Dr. Rajesh Sundaram as its technical expert on issues of validity and invalidity. Dr. Sundaram is a Professor of Electrical & Computer Engineering at Pacific Western University in San Diego, California. He was retained by Lumicore's litigation counsel, Hawthorne Vega & Cruz LLP. Dr. Sundaram's expert report on invalidity is due as Lumicore's rebuttal expert report on July 15, 2025, pursuant to the Court's scheduling order. We have not yet received Dr. Sundaram's report, and therefore the analysis herein is based solely on Lumicore's written invalidity contentions and the prior-art references themselves.
Lumicore's prior-art search was conducted by Marlowe Asset Management Patent Analytics, Inc., located at 1100 Peachtree Street NE, Suite 2200, Atlanta, Georgia 30309. Marlowe Asset Management is a well-known patent search firm that regularly supports defendants in technology-sector patent litigation. As discussed in Section VIII.B below, we recommend pursuing targeted discovery regarding Marlowe Asset Management's search methodology and internal communications with Lumicore's counsel, as those communications may reveal that Marlowe Asset Management identified the same deficiencies in the prior art that we have identified in this memorandum — specifically, the insertion-loss gap — but that Lumicore proceeded with its invalidity contentions notwithstanding those identified shortcomings.
B. Table of Prior-Art References Asserted Against the '847 Patent
The following table summarizes the six prior-art references asserted by Lumicore against the '847 Patent, including the type of reference, the effective date, the invalidity theory under which each is asserted, and the specific claims targeted.
All six references pre-date the '847 Patent's filing date of September 14, 2016, and therefore qualify as prior art under the applicable provisions of 35 U.S.C. § 102(a)(1) at least with respect to that filing date.
However, as noted above, the '847 Patent is a continuation of the '562 Patent (filed November 22, 2014). To the extent that certain '847 claims are entitled to the benefit of the earlier November 22, 2014 filing date based on adequate support in the parent '562 specification, references with effective dates between November 22, 2014 and September 14, 2016 would not qualify as prior art against those claims. This priority-date issue is particularly significant for two of the six references: Simonsen (filed September 17, 2015) and Andersson (published June 2015). The priority-date analysis is addressed in Section V.B of this memorandum.
IV. DETAILED ANALYSIS OF EACH PRIOR-ART REFERENCE
This section provides a detailed reference-by-reference analysis of each of the six prior-art references asserted by Lumicore against the '847 Patent. For each reference, we describe the content of the reference, summarize Lumicore's invalidity theory, and identify the strengths and weaknesses of Lumicore's arguments. We also provide preparation notes for Dr. Ashford's expert rebuttal report where appropriate.
A. Reference 1: Takahashi et al., "Dense WDM Silicon Photonic MUX" (March 2014) — Anticipation (§ 102)
1. Overview of the Reference
Takahashi et al. is a journal article published in IEEE Photonics Technology Letters in March 2014. The article describes a dense wavelength-division multiplexing (DWDM) silicon photonic multiplexer fabricated using an arrayed waveguide grating (AWG) architecture on a silicon-on-insulator (SOI) platform. The device reported in the article is a 16-channel AWG multiplexer with a channel spacing of 200 GHz, designed for potential use in telecommunications and data-center interconnect applications.
The authors report experimental characterization of the fabricated device, including measurements of insertion loss, channel crosstalk, and spectral uniformity. The reported insertion loss is approximately 2.8 dB per channel (averaged across all 16 channels), with some channel-to-channel variation. Crosstalk between adjacent channels is reported at approximately -15 dB. The article provides a description of the device architecture, including input waveguides, slab coupler regions, the arrayed waveguide grating, and output waveguides.
Critically, the Takahashi article describes a proof-of-concept laboratory device. There is no discussion of commercial implementation, thermal stability, or scalability to commercial manufacturing environments. The article does not discuss any specific techniques for reducing insertion loss below the reported 2.8 dB level, nor does it identify insertion loss as a target for further optimization. The article does not disclose any tapered waveguide transition regions between the slab coupler and the AWG, and it does not disclose any structures specifically designed for reducing crosstalk between adjacent channels (such as adiabatic bends or other mode-confining geometries).
2. Lumicore's Anticipation Theory
Lumicore contends that Takahashi anticipates Claims 1 and 14 of the '847 Patent by disclosing each element of the independent claims. Specifically, Lumicore maps the following elements:
• Takahashi's 16-channel AWG architecture is mapped to the "plurality of input waveguides" and "arrayed waveguide grating structure" limitations;
• Takahashi's described slab coupler regions (present on both the input and output sides of the AWG) are mapped to the "slab coupler region" limitation;
• Takahashi's output waveguide, which carries the multiplexed optical signal, is mapped to the "output waveguide" limitation;
• Takahashi's reported performance data is argued to satisfy the "low insertion loss" limitation.
On the final point — the insertion loss limitation — Lumicore's contention chart attempts to overcome the obvious deficiency in Takahashi's reported data by referencing a single data point at channel 8, where the measured insertion loss was approximately 1.9 dB. Lumicore appears to argue that this individual data point, or the performance at select channels, is sufficient to meet the "less than 1.5 dB per channel" limitation.
3. Analysis and Weaknesses
Critical Deficiency — Insertion Loss Limitation. Claim 1 of the '847 Patent unambiguously requires that the multiplexer exhibit "low insertion loss" defined as "less than 1.5 dB per channel." Takahashi reports insertion loss of approximately 2.8 dB per channel, averaged across the 16-channel device. This value is nearly double the claimed threshold. Even the most favorable individual channel data point cited by Lumicore — 1.9 dB at channel 8 — does not meet the less-than-1.5 dB requirement. The gap between Takahashi's reported performance and the claimed performance is substantial: a difference of approximately 1.3 dB on average, and 0.4 dB even at the single best channel.
This deficiency is fatal to Lumicore's anticipation theory. It is black-letter law that anticipation under 35 U.S.C. § 102 requires that a single prior-art reference disclose each and every limitation of the claimed invention, either expressly or inherently. Schering Corp. v. Geneva Pharms., Inc., 339 F.3d 1373, 1377 (Fed. Cir. 2003). There is no dispute that Takahashi does not disclose insertion loss of less than 1.5 dB per channel. Lumicore's attempt to satisfy this limitation by cherry-picking a single channel's performance (which itself does not meet the threshold) is legally and technically insufficient. The claim requires the recited performance characteristic for the multiplexer as a whole — not for a single isolated channel — and even if the limitation were construed on a per-channel basis, no individual channel in Takahashi's data achieves sub-1.5 dB performance.
Moreover, the insertion-loss gap is not a trivial distinction or a matter of routine optimization. In silicon photonic waveguide multiplexers, reducing insertion loss from the 2.8 dB range to below 1.5 dB represents a significant technological achievement that required the specific structural innovations disclosed in the '847 Patent — including the novel slab coupler geometry and the tapered waveguide transition between the slab coupler and the AWG. The magnitude of this performance gap, and the structural innovations needed to close it, are precisely the contributions of the '847 Patent to the art.
Tapered Waveguide Transition (Claim 5). Although Lumicore asserts Takahashi only for anticipation of Claims 1 and 14, it is worth noting that Takahashi does not disclose a tapered waveguide transition region between the slab coupler and the arrayed waveguide grating — the limitation added by dependent Claim 5. Lumicore's contention chart does not even attempt to map this limitation under the anticipation theory, confirming that Takahashi lacks this structural element. This absence further underscores the distance between Takahashi's proof-of-concept device and the '847 Patent's claimed invention.
Crosstalk Reduction Means (Claim 9). Takahashi does not disclose any structure corresponding to the "means for reducing crosstalk between adjacent channels" as recited in Claim 9. As noted above, Takahashi reports crosstalk levels of approximately -15 dB. While -15 dB crosstalk might be acceptable for certain narrowband laboratory demonstrations, it falls well below the performance levels required for the high-density, commercial-grade data-center interconnect applications contemplated by the '847 Patent, which typically demand crosstalk levels of -25 dB or better. Takahashi does not identify crosstalk as a problem to be solved, does not describe any structures designed to address crosstalk, and does not disclose adiabatic bends or any equivalent structure. Again, Lumicore does not assert Takahashi for anticipation of Claim 9, but this gap is relevant to the completeness of Takahashi as a prior-art reference.
Assessment. Takahashi does not anticipate any asserted claim of the '847 Patent. The reference fails on at least two independent limitations: the insertion loss threshold (2.8 dB vs. <1.5 dB) and the absence of a tapered waveguide transition (relevant to Claim 5 and, indirectly, to the structural completeness of the claimed multiplexer). The anticipation theory is the weakest element of Lumicore's invalidity case for the '847 Patent, and we do not anticipate that Lumicore's expert, Dr. Sundaram, will be able to rehabilitate this theory in his report. Risk level: Very Low.
Preparation Note for Dr. Ashford. In her rebuttal report, Dr. Ashford should: (i) clearly identify and quantify the insertion-loss gap between Takahashi (2.8 dB) and the <1.5 dB claim limitation; (ii) explain why this performance difference is technologically significant — i.e., it is not merely a matter of routine design choice but reflects a fundamental limitation of the Takahashi architecture that could not be overcome without the specific structural innovations claimed in the '847 Patent; (iii) demonstrate that the innovations disclosed in the '847 Patent (tapered waveguide transition, modified slab coupler geometry, adiabatic bend structures) are what enabled the sub-1.5 dB performance; and (iv) address and rebut any argument that Takahashi inherently discloses sub-1.5 dB performance or that a POSITA reading Takahashi would understand the reference as teaching such performance.
B. Reference 2: PCT/EP2015/071234 (Simonsen) — Obviousness (§ 103), Combined with Takahashi
1. Overview of the Reference
Simonsen is a PCT (Patent Cooperation Treaty) application filed on September 17, 2015, and published on March 23, 2016, under international publication rules. The applicant is a European entity, and the named inventor is Simonsen. The application describes a silicon photonic multiplexer that employs a modified slab coupler geometry intended to reduce insertion loss in arrayed waveguide grating-based multiplexer devices.
The Simonsen application reports simulated insertion loss values ranging from approximately 1.2 dB to 1.8 dB per channel, depending on channel count. Specifically, the 8-channel design achieves simulated insertion loss of approximately 1.2 dB, while the 16-channel design achieves simulated insertion loss of approximately 1.8 dB. These values are derived from computational electromagnetic simulations — specifically, finite-difference time-domain (FDTD) simulations — and are not based on measurements of a fabricated device. The Simonsen application does not report any experimental validation of the simulated performance, nor does it describe fabrication of a physical prototype.
The Simonsen application discloses both an 8-channel and a 16-channel multiplexer design. However, it does not disclose a tapered waveguide transition region between the slab coupler and the arrayed waveguide grating. It also does not disclose any crosstalk reduction structures corresponding to the "means for reducing crosstalk between adjacent channels" recited in Claim 9 of the '847 Patent, such as adiabatic bends or equivalent geometries.
2. Lumicore's Obviousness Theory
Lumicore contends that it would have been obvious to a person of ordinary skill in the art (POSITA) to combine Takahashi's AWG multiplexer architecture with Simonsen's modified slab coupler geometry to arrive at a silicon photonic multiplexer meeting the <1.5 dB insertion loss requirement of Claim 1. Lumicore asserts this combination against all five asserted claims — Claims 1, 3, 5, 9, and 14.
Lumicore's stated motivation to combine is that both references are in the same field of silicon photonic multiplexers and that a POSITA would have been motivated to incorporate Simonsen's slab coupler modification into Takahashi's AWG architecture in order to reduce insertion loss. Lumicore characterizes this as a straightforward combination of known elements to achieve a predictable result.
3. Analysis and Weaknesses
Insertion Loss — Reliance on Simulated Data. Even accepting Lumicore's proposed combination of Takahashi and Simonsen, the argument depends critically on Simonsen's reported insertion loss values — which are entirely simulated, not measured. In the field of silicon photonic device design, it is well recognized that simulated performance values are typically optimistic compared to the performance of fabricated devices. Simulation models do not fully account for fabrication variations, surface roughness, material impurities, waveguide sidewall imperfections, and other real-world effects that increase insertion loss in physical devices.
A POSITA would understand that Simonsen's simulated values of 1.2–1.8 dB would likely translate to higher measured values once the device was actually fabricated — potentially significantly higher. The gap between simulated and measured performance in silicon photonic AWG devices is well documented in the literature and can range from 0.5 dB to over 1.0 dB depending on the specific design and fabrication process. Applying this known simulation-to-fabrication gap to Simonsen's simulated values suggests that the actual fabricated performance would likely be in the range of 1.7–2.8 dB — substantially above the <1.5 dB threshold required by Claim 1.
This reliance on simulated data is a significant vulnerability for Lumicore's obviousness theory. Lumicore cannot establish that the combination of Takahashi and Simonsen would "predictably" yield the claimed performance when the performance data on which the argument is based has never been experimentally validated.
Tapered Waveguide Transition (Claim 5). Neither Takahashi nor Simonsen discloses the specific tapered waveguide transition recited in Claim 5. Lumicore's contention chart for Claim 5, when relying on the Takahashi + Simonsen combination, either leaves the tapered waveguide transition limitation unmapped or cross-references Bergmann (Reference 3) for this element. This means that for Claim 5, Lumicore actually requires a three-reference combination — Takahashi + Simonsen + Bergmann — to account for all claim limitations. The addition of a third reference introduces additional motivation-to-combine requirements and increases the hindsight reconstruction concern, as discussed below in the analysis of Bergmann.
Crosstalk Reduction Means (Claim 9). Similarly, neither Takahashi nor Simonsen discloses structures corresponding to Claim 9's "means for reducing crosstalk between adjacent channels," as construed under 35 U.S.C. § 112(f) with corresponding structure of adiabatic bends with specific geometry. Again, Lumicore relies on Bergmann for this limitation, requiring a three-reference combination for Claim 9 as well.
Motivation to Combine. Lumicore's stated motivation to combine — that "both references relate to silicon photonic multiplexers and a POSITA would combine them to reduce loss" — is a generic, conclusory assertion that is insufficient under Federal Circuit precedent. A legally adequate motivation to combine requires more than the observation that two references are in the same general field. The proponent must identify a specific teaching, suggestion, or motivation in the references, the knowledge of a POSITA, or the nature of the problem to be solved that would have led a POSITA to combine the references. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398, 418 (2007); In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006).
Lumicore's motivation-to-combine argument is further weakened by the fact that both Takahashi and Simonsen were published and publicly available for over a year before the '847 Patent's filing date, yet no one in the field — including the authors of these references and other groups working on silicon photonic multiplexers — achieved the combination and performance claimed in the '847 Patent during that period. This "failure of others" evidence cuts strongly against obviousness and supports the inference that the combination was not as straightforward as Lumicore suggests.
Assessment. The Takahashi + Simonsen combination is insufficient for Claims 5 and 9 without the addition of Bergmann (creating a three-reference combination), and the motivation to combine is weak and conclusory for all claims. The reliance on simulated data rather than measured performance is a significant vulnerability that Dr. Ashford can effectively exploit in her rebuttal report. Risk level: Low. However, this combination is marginally stronger than the anticipation theory because it at least identifies a reference (Simonsen) that addresses the insertion loss limitation in some form, even if only through simulated data. We classify the overall risk as moderate-to-low for Claim 1 and low for the remaining claims.
C. Reference 3: U.S. Patent No. 8,412,007 (Bergmann) — Obviousness (§ 103), Combined with Takahashi and/or Simonsen
1. Overview of the Reference
Bergmann is U.S. Patent No. 8,412,007, issued on April 2, 2013. The patent discloses a photonic integrated circuit comprising waveguide structures including tapered transitions and adiabatic bends. The technology context of Bergmann is critical: the patent is directed to a III-V semiconductor platform, specifically indium phosphide (InP), not to silicon photonics. InP is a fundamentally different material system from silicon-on-insulator (SOI), with different refractive index contrast, waveguide geometry, fabrication processes, and optical mode characteristics.
Bergmann describes techniques for reducing crosstalk in waveguide arrays using specific bend geometries, including adiabatic bends that gradually change the waveguide direction while maintaining optical mode confinement. The patent also discloses tapered waveguide transitions that are used to match mode sizes between different waveguide cross-sections. However, Bergmann does not disclose a waveguide multiplexer per se — the patent is focused on individual waveguide components and design techniques applicable to photonic integrated circuits generally, rather than on complete multiplexer architectures.
2. Lumicore's Obviousness Theory
Lumicore argues that Bergmann supplies the tapered waveguide transition (relevant to Claim 5) and the crosstalk reduction means (relevant to Claim 9) that are missing from Takahashi and Simonsen. Lumicore's theory is that a POSITA would have combined Bergmann's waveguide design techniques with the Takahashi/Simonsen multiplexer architecture to arrive at a device meeting the limitations of Claims 1, 5, and 9.
This requires either a two-reference combination (Takahashi + Bergmann) or, more commonly in Lumicore's contention charts, a three-reference combination (Takahashi + Simonsen + Bergmann). The three-reference combination is required when Lumicore relies on Simonsen for the insertion-loss improvement and Bergmann for the structural elements of Claims 5 and 9.
3. Analysis and Weaknesses
Different Technology Platform. The most fundamental weakness of the Bergmann combination is the platform mismatch. Bergmann is directed to III-V photonics — specifically, InP-based photonic integrated circuits. The '847 Patent is directed to silicon photonics — specifically, silicon-on-insulator (SOI) platform devices. These material systems differ in critical respects that would be well understood by a POSITA:
• Refractive index contrast: SOI waveguides have a much higher refractive index contrast (Δn ≈ 2.0 for Si/SiO₂) compared to InP waveguides (Δn ≈ 0.3 for InGaAsP/InP). This difference profoundly affects waveguide dimensions, mode confinement, bending radii, and the design of tapered transitions.
• Waveguide geometry: Due to the high index contrast, SOI waveguides are typically sub-micron in cross-section (e.g., 500 nm × 220 nm), while InP waveguides are several micrometers in dimension. Tapered transitions and adiabatic bends designed for InP waveguides cannot be directly scaled to SOI dimensions without complete redesign.
• Fabrication constraints: SOI devices are fabricated using CMOS-compatible processes (e.g., deep-UV lithography, dry etching), while InP devices use different fabrication techniques (e.g., wet etching, epitaxial regrowth). The fabrication process affects the achievable waveguide geometries and surface quality, which in turn affect insertion loss and crosstalk.
A POSITA with expertise in silicon photonics would recognize that Bergmann's InP-based design techniques cannot be straightforwardly transplanted to an SOI AWG multiplexer. The physics governing optical mode propagation, coupling, and crosstalk are fundamentally different at the dimensional and material-property scales involved. Lumicore will bear the burden of establishing that Bergmann's tapered transition and adiabatic bend techniques are "platform-agnostic" — i.e., applicable regardless of the underlying material system — and this is a contested technical point that we believe Dr. Ashford can effectively rebut.
Bergmann Does Not Disclose a Multiplexer. Bergmann's patent is directed to individual waveguide components — tapered transitions, adiabatic bends, and waveguide routing elements — not to a complete multi-channel wavelength multiplexer. There is a significant conceptual gap between knowing how to design an individual waveguide taper or bend (Bergmann's contribution) and knowing how to integrate such components into a multi-channel AWG multiplexer to achieve specific system-level performance characteristics like sub-1.5 dB insertion loss and low crosstalk (the '847 Patent's contribution). Combining Bergmann's component-level teachings with Takahashi's system-level device requires a leap that a POSITA would not make without the benefit of hindsight — knowing in advance that the specific combination of components would yield the desired system-level performance.
The Federal Circuit has cautioned against this type of hindsight reconstruction in the obviousness analysis. In re Dembiczak, 175 F.3d 994, 999 (Fed. Cir. 1999) ("It is impermissible to use the patent in suit as a blueprint for piecing together the prior art to defeat the patent."). Here, Lumicore's theory essentially requires a POSITA to identify the specific structural features of the '847 Patent (tapered slab-coupler-to-AWG transition, adiabatic bends) as solutions to the insertion loss and crosstalk problems, and then to locate those features in a component-level InP patent — a search that would only be undertaken with knowledge of the claimed invention.
Three-Reference Combination. For Claims 5 and 9, Lumicore requires Takahashi + Simonsen + Bergmann — a three-reference combination spanning two technology platforms (SOI and InP) and two levels of device hierarchy (system-level multiplexer and component-level waveguide). Courts and the Federal Circuit have recognized that as the number of references required for a combination increases, the argument for non-obviousness strengthens, particularly when the references come from different technology domains. In re Kubin, 561 F.3d 1351, 1358 (Fed. Cir. 2009); see also Wyers v. Master Lock Co., 616 F.3d 1231, 1240 (Fed. Cir. 2010). While there is no per se rule against multi-reference combinations, the burden on the party asserting obviousness to establish a credible motivation to combine increases with each additional reference, and the inference of hindsight becomes stronger.
Assessment. The Bergmann combination is the weakest of Lumicore's obviousness theories for the '847 Patent. The cross-platform transplant from InP to SOI, the component-versus-system gap, and the requirement for a three-reference combination all create significant vulnerabilities that we are confident Dr. Ashford can effectively exploit. Risk level: Very Low.
Note on Overlap with '562 Patent: Bergmann is also cited by Lumicore against the '562 Patent (see companion memo dated November 5, 2024). The analysis of Bergmann's platform mismatch set forth above applies equally to both patents. The case team should ensure consistency between Dr. Ashford's treatment of Bergmann in the '847 Patent and '562 Patent rebuttal reports.
D. Reference 4: Okamoto, "Fundamentals of Optical Waveguides" (2006 Textbook) — Obviousness (§ 103), Background Knowledge
1. Overview of the Reference
Okamoto's "Fundamentals of Optical Waveguides" is a standard textbook on optical waveguide theory, published in its second edition in 2006. The textbook is widely used in graduate-level courses in photonics and optical engineering and is a well-known reference work in the field. The textbook covers a broad range of waveguide theory topics, including mode theory, coupled-mode theory, beam propagation methods, and the design of various integrated optical devices.
Relevant to this case, the textbook includes a chapter on arrayed waveguide grating (AWG) design principles. This chapter provides a general discussion of AWG theory, including the operating principles of slab couplers, the phase relationships in waveguide arrays, insertion loss mechanisms, channel crosstalk, and spectral response characteristics. The discussion is at a theoretical and pedagogical level, covering the fundamental physics of AWG operation across multiple material platforms — including glass, polymer, and semiconductor waveguides. The textbook does not disclose a silicon photonic multiplexer specifically, nor does it provide design recipes or fabrication guidance for achieving specific performance levels on any particular platform.
Critically for purposes of the invalidity analysis: Okamoto does not disclose specific insertion loss values below 1.5 dB for any silicon photonic implementation. The textbook discusses insertion loss as a general performance parameter and identifies the physical mechanisms that contribute to insertion loss in AWG devices (e.g., mode mismatch at slab coupler interfaces, propagation loss in waveguide arrays, phase errors), but it does not teach how to achieve the specific sub-1.5 dB performance level claimed in the '847 Patent. The textbook also does not disclose tapered waveguide transitions for AWG multiplexers or crosstalk reduction structures as recited in Claim 9.
2. Lumicore's Theory
Lumicore cites Okamoto not as a primary invalidating reference but as evidence of the background knowledge available to a POSITA. The theory is that Okamoto establishes that AWG multiplexer design principles were well known in the art, thereby lowering the threshold of inventiveness required for the '847 Patent claims. Combined with Takahashi, Lumicore argues that a POSITA with knowledge of the Okamoto textbook would have known how to optimize the Takahashi design to achieve the performance levels claimed in the '847 Patent.
Lumicore asserts this combination against Claims 1, 3, and 14.
3. Analysis and Weaknesses
Textbook as Background Reference. We do not dispute that Okamoto is a well-known reference in the field, and we expect the Court to accept that a POSITA would be familiar with the general principles described in the textbook. However, familiarity with general AWG design principles does not render the specific claims of the '847 Patent obvious. The textbook discusses waveguide theory at a level of abstraction that does not teach the specific combination of structural features and performance characteristics recited in the claims. Knowing the general theory of how AWGs work is not the same as knowing how to design a specific silicon photonic AWG multiplexer that achieves sub-1.5 dB insertion loss — if it were, the sub-1.5 dB performance level would have been achieved long before OWS's invention, given that Okamoto's textbook was published in 2006.
"Mere Optimization" Argument Is Legally Vulnerable. Lumicore's reliance on Okamoto appears designed to support an argument that achieving sub-1.5 dB insertion loss was merely a matter of "routine optimization" of known design parameters. This argument is legally vulnerable under Federal Circuit precedent, which holds that the "mere optimization" theory applies only when the parameter at issue is a "result-effective variable" — i.e., a variable that a POSITA would have known to vary, and that the claimed range was achievable through routine experimentation. In re Applied Materials, Inc., 692 F.3d 1289, 1297 (Fed. Cir. 2012); In re Aller, 220 F.2d 454, 456 (CCPA 1955).
Here, the '847 Patent claims are not directed to routine optimization of a single known parameter. Achieving sub-1.5 dB insertion loss on an SOI platform required specific structural innovations — including the tapered waveguide transition between the slab coupler and the AWG, the modified slab coupler geometry, and the adiabatic bend structures — that go beyond varying a single design parameter within a known range. This is confirmed by the prosecution history, in which the examiner allowed the claims over the prior art precisely because the combination of the insertion-loss threshold and the structural limitations distinguished the claims from the prior art. The examiner's finding, while not dispositive, is entitled to a presumption of correctness under 35 U.S.C. § 282.
Assessment. Okamoto adds minimal incremental force to Lumicore's invalidity case for the '847 Patent. It establishes general background knowledge but does not supply any missing claim limitation. The textbook does not disclose the specific insertion loss values, structural features, or platform-specific design techniques claimed in the '847 Patent. Okamoto is useful to Lumicore only as a supporting reference in combination arguments, and its contribution is largely cumulative of what any POSITA would already know. Risk level: Very Low as an independent risk; marginal supporting value in combination arguments.
E. Reference 5: U.S. Patent Application Pub. No. 2015/0XXX (Andersson) — Obviousness (§ 103), Combined with Takahashi
1. Overview of the Reference
Andersson is a published U.S. patent application with a publication date of June 2015. The application discloses a multi-channel photonic device designed for telecommunications applications. Andersson describes devices with channel counts of 4, 8, and 16 channels and discusses techniques for reducing crosstalk between adjacent channels using modified waveguide spacing and mode confinement techniques.
A critical distinction is that Andersson's device is fabricated on a silicon nitride (SiN) platform, not on a silicon-on-insulator (SOI) platform. While both SiN and SOI are "silicon-compatible" platforms in the sense that they can be fabricated in CMOS-compatible foundries, they have significantly different optical properties. SiN has a lower refractive index contrast than SOI (Δn ≈ 0.5 for SiN/SiO₂ vs. Δn ≈ 2.0 for Si/SiO₂), resulting in larger waveguide dimensions, different mode profiles, and different design constraints for components such as bends, tapers, and couplers.
Andersson does not disclose insertion loss values meeting the <1.5 dB requirement of the '847 Patent claims. The application does not disclose a tapered waveguide transition region between a slab coupler and an AWG structure.
2. Lumicore's Theory
Lumicore combines Andersson with Takahashi to address Claim 3 (the 8-channel limitation) and Claim 9 (the crosstalk reduction means). Lumicore argues that Andersson's disclosed channel counts (including 8 channels) and crosstalk reduction techniques (modified waveguide spacing and mode confinement) would be combined with Takahashi's AWG architecture by a POSITA to arrive at the claimed invention.
3. Analysis and Weaknesses
Platform Mismatch. As with Bergmann, Andersson presents a platform mismatch issue — though the mismatch here is between SiN and SOI rather than between InP and SOI. While the SiN-to-SOI gap is arguably smaller than the InP-to-SOI gap (because SiN and SOI are both fabricated in CMOS-compatible processes), the two platforms still have significantly different optical properties. The refractive index contrast, waveguide dimensions, and mode confinement characteristics differ sufficiently that crosstalk reduction techniques optimized for SiN waveguides would not straightforwardly transfer to SOI waveguides without substantial redesign.
In particular, Andersson's crosstalk reduction approach relies on modified waveguide spacing — i.e., increasing the physical distance between adjacent waveguides to reduce evanescent-field coupling. This approach is feasible in the SiN platform, where waveguide dimensions are larger and device footprints are accordingly less constrained. In the SOI platform, where the high refractive index contrast enables much smaller waveguide dimensions and tighter integration, simply increasing waveguide spacing may not be a viable or desirable solution because it would increase the device footprint and negate one of the primary advantages of silicon photonics (i.e., high integration density).
Claim 3 — 8-Channel Limitation. Lumicore's argument that Andersson supplies the 8-channel limitation for Claim 3 is weak and unnecessary because Takahashi already discloses a 16-channel device. Reducing the channel count from 16 to 8 is a trivial modification that does not require a separate prior-art reference — a POSITA designing an AWG multiplexer would routinely select the channel count based on application requirements. However, this point is largely academic because Claim 3 depends on Claim 1, and therefore all of the deficiencies identified for Claim 1 carry through to Claim 3. Even if Lumicore can establish the 8-channel limitation, it cannot establish the <1.5 dB insertion loss limitation without relying on the other combinations analyzed above.
Claim 9 — Crosstalk Reduction. Andersson's crosstalk reduction technique — modified waveguide spacing — is structurally different from the "means for reducing crosstalk between adjacent channels" recited in Claim 9 of the '847 Patent. As determined in our Means-Plus-Function Analysis Memorandum (October 5, 2024), the corresponding structure for Claim 9 is adiabatic bends with specific geometry disclosed in the '847 Patent specification. Andersson's approach of modifying waveguide spacing is neither the same structure as adiabatic bends nor a structural equivalent. The two approaches reduce crosstalk through different physical mechanisms: waveguide spacing reduces crosstalk by increasing the physical distance between waveguides (and thereby reducing evanescent-field overlap), while adiabatic bends reduce crosstalk by maintaining mode confinement through gradual waveguide curvature changes that prevent mode coupling at bend points.
Under means-plus-function analysis, a prior-art reference must disclose either the identical structure or a structural equivalent that performs substantially the same function, in substantially the same way, to achieve substantially the same result. Odetics, Inc. v. Storage Tech. Corp., 185 F.3d 1259, 1267 (Fed. Cir. 1999). Andersson's waveguide spacing modification does not meet this standard.
Assessment. Andersson adds limited value to Lumicore's invalidity case. The platform mismatch and the structural differences in the crosstalk reduction approach are significant weaknesses. The 8-channel limitation is easily met by other means and does not independently strengthen Lumicore's position. Risk level: Low.
F. Reference 6: Chen & Park, "Low-Loss AWG Design on SOI" (ECOC 2014 Conference Paper) — Obviousness (§ 103), Combined with Takahashi
1. Overview of the Reference
Chen & Park is a conference paper presented at the European Conference on Optical Communication (ECOC) in September 2014. The paper describes an arrayed waveguide grating (AWG) multiplexer design fabricated on a silicon-on-insulator (SOI) platform with a focus on reducing insertion loss. The authors report measured insertion loss of approximately 1.8–2.2 dB per channel, representing an improvement over earlier SOI AWG demonstrations but still above the <1.5 dB threshold claimed in the '847 Patent.
Chen & Park discuss the use of optimized slab coupler geometry and waveguide width tapering as techniques to reduce insertion loss. However, the waveguide width tapering described in the paper occurs at the input/output waveguide interfaces — i.e., at the points where the input and output waveguides connect to the slab coupler — and serves the purpose of mode-size matching to improve coupling efficiency at those interfaces. This is distinct from the "tapered waveguide transition region between the slab coupler and the arrayed waveguide grating" recited in Claim 5 of the '847 Patent, which is located at a different position in the device (at the interface between the slab coupler and the arrayed waveguide grating, not at the input/output waveguide interfaces) and serves a different purpose (reducing coupling loss at the slab-coupler-to-AWG interface, not input/output mode matching).
The Chen & Park paper does not explicitly disclose insertion loss below 1.5 dB per channel. The best reported performance is approximately 1.8 dB for the lowest-loss channels, with most channels exhibiting loss in the 2.0–2.2 dB range.
2. Lumicore's Theory
Lumicore combines Chen & Park with Takahashi, arguing that application of Chen & Park's loss-reduction techniques (optimized slab coupler geometry and waveguide width tapering) to Takahashi's AWG design would render Claims 1, 5, and 14 obvious. Lumicore argues that the combination of loss-reduction techniques from both references would predictably yield sub-1.5 dB insertion loss.
Lumicore's motivation-to-combine argument is that both references address the same problem — reducing insertion loss in SOI-based AWG multiplexers — and that a POSITA would naturally combine the best techniques from both references.
3. Analysis and Weaknesses
Insertion Loss Gap Persists. Chen & Park report measured insertion loss of 1.8–2.2 dB per channel — an improvement over Takahashi's 2.8 dB but still above the <1.5 dB threshold required by Claim 1. Combining Takahashi (2.8 dB) with Chen & Park's loss-reduction techniques might plausibly bring the insertion loss value closer to 1.8 dB but does not establish that a POSITA would have achieved sub-1.5 dB performance. The gap between Chen & Park's best results (1.8 dB) and the <1.5 dB claim limitation — a difference of at least 0.3 dB — is significant in the context of silicon photonic waveguide design, where each incremental 0.1 dB reduction in insertion loss becomes progressively more difficult to achieve. The "last mile" from 1.8 dB to below 1.5 dB required structural innovations beyond what Chen & Park or Takahashi disclosed.
There is no evidence in either reference that the sub-1.5 dB threshold was within the reach of routine optimization. Neither Chen & Park nor Takahashi identifies the <1.5 dB level as a design target or suggests that further incremental improvements using their disclosed techniques would reach that level. To the contrary, Chen & Park's paper discusses the challenges of further reducing insertion loss in SOI AWG devices and identifies several loss mechanisms that are difficult to mitigate with the techniques available at the time.
Tapered Waveguide Transition (Claim 5). Chen & Park disclose "waveguide width tapering" in a general sense, but the tapering described in their paper is structurally and functionally distinct from the tapered waveguide transition recited in Claim 5 of the '847 Patent. Specifically:
• Location: Chen & Park's tapering occurs at the input/output waveguide interfaces (where the single-mode waveguides connect to the slab coupler). The '847 Patent's tapered transition is located between the slab coupler and the arrayed waveguide grating (on the opposite side of the slab coupler from the input/output waveguides).
• Purpose: Chen & Park's tapering is designed for mode-size matching — i.e., gradually expanding the waveguide width to match the slab coupler's mode field profile. The '847 Patent's tapered transition is designed to reduce coupling loss at the slab-coupler-to-AWG interface by providing a gradual transition in waveguide geometry that minimizes reflections and scattering.
• Design parameters: Because the two tapers are at different locations and serve different functions, their design parameters (taper length, width ratio, profile shape) are necessarily different. A POSITA would not conflate the two.
Lumicore's mapping of Chen & Park's waveguide tapering to the Claim 5 limitatio

CALDERWOOD & HARKNESS LLP

PRIVILEGED AND CONFIDENTIAL

I. EXECUTIVE SUMMARY

II. BACKGROUND ON THE '847 PATENT

III. OVERVIEW OF LUMICORE'S INVALIDITY CONTENTIONS

IV. DETAILED ANALYSIS OF EACH PRIOR-ART REFERENCE

Record

Parties

Dates

Amounts