Dario Lo Cascio’s Post

📐 #TBT | Silicon Modeling: Can we simplify CMOS pad extraction? For today’s #ThrowbackThursday, I’m sharing a classic paper we published in Microwave and Optical Technology Letters (MOTL) that addresses a persistent challenge in integrated circuit design: dealing with lossy silicon substrates. The Challenge: “Can two simple uniform transmission lines of different lengths accurately extract the electrical equivalent circuit model of CMOS pads on lossy Si substrates?” 🤔 Our Approach: We developed a practical method for modeling CMOS pads that can be represented by symmetrical and reciprocal ABCD matrices, which are then converted into a T-network representation using Z-parameters. The Impact: The proposed extraction method enabled highly accurate electrical modeling and experimental validation of CMOS pads over a wide frequency range (0.4-20 GHz). It's a great reminder that the mathematical elegance of transmission lines remains a cornerstone in modern high-frequency design. Saavedra-Gomez, H.J., Loo-Yau, J.R., Hernandez, J.A.R., Moreno, P. and del Valle-Padilla, J.L. (2013), A Simple and Reliable Method to Extract The Electrical Equivalent Circuits of CMOS pads. Microw. Opt. Technol. Lett., 55: 3033-3037. https://lnkd.in/gu7QXXUK 👉 To our alumni currently dealing with pad parasitics in the semiconductor industry: Are you still using similar de-embedding approaches in your characterizations? Follow our page to stay updated with our historical milestones, current RF/microwave measurements, and upcoming innovations. #ThrowbackThursday #RFEngineering#RF #Microwave #CMOS #Semiconductors #Cinvestav #MOTL #CircuitDesign #ICDesign

Technology Spotlight: Understanding Laser Diodes From the high-speed data streams in telecommunications to the precision tools in our workshops and medical facilities, Laser Diodes are the silent heroes of modern technology. But how do they differ from standard LEDs? At their core, laser diodes are semiconductor devices that take light generation to a new level. While an LED produces incoherent light, a laser diode utilises stimulated emission to generate a coherent, monochromatic beam. How they work: * Energy Injection: Current is passed into a PN junction, pushing electrons into a higher energy state. * Spontaneous Emission: As electrons recombine with holes, they release energy as photons. * Stimulated Emission (The "Lasing" Effect): By fabricating partially reflecting surfaces at the ends of the structure, trapped photons stimulate other excited electrons to recombine faster. This cascades into the creation of more photons—all with the exact same frequency and phase—resulting in that powerful, focused laser beam. Why they are indispensable: * Solid-State Reliability: No fragile glass tubes or complex setups; they are compact, rugged, and built for modern electronics. * Performance: Capable of being focused to a diffraction-limited spot, which is essential for high-density optical storage and precision scanning. * Versatility: By changing the compound semiconductor materials, engineers can tune the output wavelength from visible blue to far infrared. Whether you're working on optical communications, industrial sensing, or consumer electronics, understanding the characteristics—like the L/I (Light/Current) curve and the critical importance of temperature control—is key to successful circuit design. Check out the full technical breakdown on Electronics Notes: https://lnkd.in/eQGsCfKK #ElectronicsEngineering #Photonics #LaserTechnology #Semiconductors #CircuitDesign #EngineeringFundamentals #Innovation #TechTrends #ElectronicComponents

The Thermal Wall: How Optical Interconnects Are Tackling the 30W+ Challenge in the AI Era As AI clusters push optical interconnects from 800G to 1.6T and 3.2T, we are facing a fundamental physical limit: The Thermal Wall. Packing 25W–30W+ into a compact form factor elevates surface heat flux to levels comparable to high-performance CPUs. Managing thermal dissipation without compromising bit error rates (BER) or module lifespan is now a core industry priority. Here is a breakdown of how the optical ecosystem is tackling this challenge, from current conduction paths to architectural paradigms: 1. Current Baseline: Advanced Conduction & Component Cooling Optimized Thermal Paths: Minimizing multi-interface thermal resistance via ultra-thin TIM1 (internal) and wear-resistant TIM2 (external). Targeted Heat Dissipation: Micro Vapor Chambers (Micro VC) for DSP hot spot spreading, combined with Micro-TECs for precise EML laser temperature control. Liquid-Cooling Readiness: High-integrity sealing and anti-corrosion packaging for immersion and cold-plate environments. 2. Core Bottlenecks: Cumulative Thermal Resistance Long Thermal Chains: Multiple interfacial boundary layers between chips, module shells, and switch heat sinks limit heat extraction speed. The TEC Vicious Cycle: Rising ambient temperatures force TECs to draw more current, adding internal thermal load. 3. The Future Roadmap: Architecture & Material Revolutions Short-to-Mid Term (800G / 1.6T): Linear & Direct Drive (LPO / TRO): Eliminating or simplifying internal DSPs removes up to 30%–50% of module power consumption. External Laser Sources (ELS): Thermally decoupling sensitive laser diodes from high-power host processing environments. Mid-to-Long Term (1.6T / 3.2T): Co-Packaged Optics (CPO): Integrating optical engines directly onto the ASIC substrate—drastically shortening electrical traces and enabling shared, chip-level liquid cold plates. Advanced Material Innovations: Diamond Heat Spreaders & Liquid Metal TIMs: Leveraging ultra-high thermal conductivity to break interfacial limits. TFLN & Silicon Photonics: High-efficiency electro-optic materials that curb heat generation at the source. #OpticalCommunications #1_6T #OpticalModules #CPO #LPO #SiliconPhotonics #ThermalManagement #DataCenter #AIInfrastructure #HardwareEngineering

𝗛𝗶𝗴𝗵 𝗰𝗼𝗻𝗳𝗼𝗿𝗺𝗮𝗹𝗶𝘁𝘆. 𝗛𝗶𝗴𝗵 𝘁𝗵𝗿𝗼𝘂𝗴𝗵𝗽𝘂𝘁. 𝗛𝗶𝗴𝗵 𝗶𝗺𝗽𝗮𝗰𝘁. Following our previous #PhotonicFriday posts, we are closing this machine series with the 𝗟𝗘𝗬𝗕𝗢𝗟𝗗 𝗢𝗣𝗧𝗜𝗖𝗦 𝗔𝗟𝗗 𝟭𝟮𝟬𝟬. The ALD 1200 brings 𝗦𝗽𝗮𝘁𝗶𝗮𝗹 𝗔𝘁𝗼𝗺𝗶𝗰 𝗟𝗮𝘆𝗲𝗿 𝗗𝗲𝗽𝗼𝘀𝗶𝘁𝗶𝗼𝗻 to advanced photonics and semiconductor applications, combining 𝗽𝗿𝗲𝗰𝗶𝘀𝗲 𝗰𝗼𝗻𝗳𝗼𝗿𝗺𝗮𝗹 𝗰𝗼𝗮𝘁𝗶𝗻𝗴 with 𝗵𝗶𝗴𝗵 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆. Unlike conventional ALD, 𝗦𝗽𝗮𝘁𝗶𝗮𝗹 𝗔𝗟𝗗 separates the process gases in space rather than only in time. This enables 𝗵𝗶𝗴𝗵 𝗰𝗼𝗻𝗳𝗼𝗿𝗺𝗮𝗹𝗶𝘁𝘆 on 𝗰𝗼𝗺𝗽𝗹𝗲𝘅 𝟯𝗗 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲𝘀 while supporting 𝗶𝗻𝗰𝗿𝗲𝗮𝘀𝗲𝗱 𝘄𝗮𝗳𝗲𝗿 𝘁𝗵𝗿𝗼𝘂𝗴𝗵𝗽𝘂𝘁. For photonic integrated circuits, this is especially relevant for 𝗦𝗶𝗢𝟮 𝗰𝗹𝗮𝗱𝗱𝗶𝗻𝗴 𝗹𝗮𝘆𝗲𝗿𝘀, where precise coating quality can influence optical performance. Coupling this with high conformality, new and more challenging structures can now be produced. 𝗧𝗵𝗲 𝗔𝗟𝗗 𝟭𝟮𝟬𝟬 𝘀𝘂𝗽𝗽𝗼𝗿𝘁𝘀: 🔹 Conformal coating of deep trenches and complex structures  🔹 Multi-wafer coating for increased WPH   🔹 Superior refractive index control to optimize light modes in LN waveguides   🔹 Reduced air gaps that can add optical loss   🔹 Passivation layers to help protect PIC performance   🔹 Material flexibility for single layers, multilayers, and doped layer stacks  Because in photonics, light performance depends on more than the design of the waveguide. 𝗧𝗵𝗮𝘁 𝗶𝘀 𝘁𝗵𝗲 𝗽𝗼𝘄𝗲𝗿 𝗼𝗳 𝗟𝗘𝗬𝗕𝗢𝗟𝗗 𝗢𝗣𝗧𝗜𝗖𝗦 𝗔𝗟𝗗 𝟭𝟮𝟬𝟬. #PhotonicsFriday #Photonics #AtomicLayerDeposition #ALD #BühlerLeyboldOptics

At Image Sensing Exhibition 2026, Hiroshi Fukui, Technical Advisor for Transmission System Technology at Mimaki Electronic Components Co., Ltd., presented a “Light Output Camera” product concept for machine vision applications. Mimaki Electronic Components, the parent company of OptoHub, has deep expertise in optical transmission technologies. OptoHub builds on this technical foundation to bring optical interface solutions into practical imaging and industrial applications. The concept integrates an E/O conversion module into the camera architecture, enabling optical signal output through fiber. This can help support: ✅Longer transmission distance ✅Higher-speed signal transfer ✅EMI-Resistant Optical Transmission ✅Reduced heat generation by eliminating the FPGA ✅More flexible machine vision system design As imaging systems continue to evolve, optical transmission can become a key enabling technology for next-generation machine vision. OptoHub will continue to explore practical optical transmission solutions for advanced imaging applications. #MachineVision #IndustrialImaging #OpticalCommunication #FiberOptics #OpticalTransmission #OptoHub

Most people think crystal defects are flaws to eliminate. At the nanoscale, they can become design features. In colloidal nanocrystal superlattices, the optical response is not determined only by individual particle size or composition. When defects become ordered across the lattice, they can redirect coupling between neighboring nanocrystals and noticeably change scattering, structural color, and even polarization behavior. That matters for anyone building optical coatings, sensors, photonic materials, or self-assembled devices: the “imperfections” can be part of the functionality. It's a strong reminder that nanoscale performance often depends as much on arrangement and defect architecture as on chemistry alone. Characterizing order, disorder, and interfaces carefully is where the real story begins. #Nanocrystals #SelfAssembly #OpticalMaterials #MaterialsCharacterization #Nanophotonics

In photonics, one of the hardest part is keeping performance when you scale. Most integrated photonics platforms force you to choose. You can have a material that delivers exceptional optical performance, but fights you the moment you try to manufacture it in volume. Or you can have something easy to produce, that quietly gives up performance along the way. For years, that trade-off shaped what was actually possible to bring to market.   When we started building Oxyn at PhotonPath, we didn't treat it as a single device. We treated it as a platform, the foundation everything else would stand on. And a platform has to hold two things at once: the performance customers need, and the repeatability that lets you ship at scale.   So the choices were deliberate.   Oxyn blends the strengths of Silicon Nitride and traditional Planar Lightwave Circuit technologies. Ultra-low-loss waveguides, at 0.1 dB/cm. High optical power handling. Tight bending radius, down to 300 µm, so designs stay compact. Low coupling losses to fiber. Standard waveguides that are polarization-insensitive, which removes a whole layer of complexity from the system around them. Efficient thermal actuators for precise tuning and reconfigurability.   Each of those is a performance number. But the real point is what they add up to: fewer compromises pushed downstream, into packaging, testing, and the customer's own system.   That's the part the industry tends to underweight. A platform isn't judged only by its best measurement in a lab. It's judged by how much of that performance survives the trip to volume manufacturing, wafer after wafer, with the same quality every time. Performance gets the attention, while repeatability is what turns a technology into infrastructure.   Oxyn was built for both, on purpose. Because in this industry, the platform you choose early decides the ceiling you hit later.

More optical power does not always mean better photodetector performance. In our recent work on lateral Ge-on-Si waveguide photodetectors, we used a three-dimensional multiphysics model to investigate how increasing optical input power affects the electro_optical bandwidth. The central question is simple: When does a photodetector stop behaving like a linear component? At high optical power, the generated carrier population can modify the internal electric field, change transport conditions and reduce the frequency response. A model that treats optical generation and carrier transport as separate problems may miss part of this interaction. This is why I find multiphysics TCAD Simulation especially useful. A model should not only reproduce device behaviour. It should help identify which physical mechanism limits performance, which operating conditions activate that limitation and which design parameter should be changed next. For industrial design, this distinction matters. Optimising low-power bandwidth may not be sufficient when the final application requires high optical input, large photocurrent and reliable operation across different bias conditions. Can simulation help define the useful operating region before the next fabrication and measurement loop? The goal is not to replace intuition. The goal is to give intuition a better dashboard. Call to action: Are you working on high-speed photodetectors, silicon photonics or TCAD models under high optical power? I would be happy to discuss modeling approaches, validation strategies and possible academic or industrial collaborations. #Photonics #TCAD #Photodetectors #SiliconPhotonics #GeOnSi #MultiphysicsModeling #SemiconductorDevices #DeviceModeling #ResearchToIndustry

⁉️ More laser power or less optical loss? When a photonic module doesn't meet its optical power budget, the first instinct is often straightforward: 🚨 Increase the laser power While this may compensate for excess optical loss, it is rarely the most efficient solution. Higher laser power can lead to: 🔷 Reduced laser lifetime 🔷 Increased power consumption 🔷 More heat generation within the package 🔷 Additional cooling requirements 🔷 Thermal hotspots that can affect neighboring photonic circuits on the PIC These challenges can be managed, but often at the expense of additional complexity, cost, and design effort. Before increasing laser power, it's worth asking: ❓ Where is the optical loss actually coming from? In many photonic systems, avoidable losses originate from the optical interface itself, such as: 🔶 Connector losses 🔶 Fiber selection and mode-field mismatch 🔶 Unnecessary fiber length 🔶 Fiber-to-PIC alignment accuracy 🔶 Fiber array core positioning accuracy Optimizing these elements reduces optical loss at its source, allowing the available laser power to be used more effectively. At MicroAlign, we focus on one of these key contributors: fiber array accuracy. By actively aligning every individual fiber, we achieve <100 nm core positioning accuracy, resulting in lower and more uniform fiber-to-chip coupling losses. 💡 The result? ✔ Higher system efficiency ✔ Lower power requirements ✔ Improved channel-to-channel consistency ✔ Better sensing performance and dynamic range ✔ More design margin for the entire photonic system The best photon is the one you never lose. Before adding more optical power, ensure you're making the most of the power you already have. Interested in learning more about our low loss fiber arrays? Download our presentation to explore MicroAlign’s active-alignment technology at https://lnkd.in/eQ4ChrQK I would be happy to discuss how high-accuracy fiber arrays can improve the optical budget of your next photonic product. #IntegratedPhotonics #FiberArrays #OpticalPackaging #OpticalNetworks #MicroAlign

I'm writing a multipart series on Lasers for co-packaged/near-packaged optics. There is a lot of FUD around this, with few people really understanding how laser with 100s of milliwatts of output power is really engineered. Part 1 deals with the basics of InP Distributed Feedback Lasers https://lnkd.in/ditj_E2K Part 2 deals with how JDSU/Lumentum engineered a 400mW laser, where the design tradeoffs lie, and how industry competitors are positioned. https://lnkd.in/d_taVj7j