🔬 optics

Inverse-Designed High-Q/V Silicon Nitride Photonic Crystal Cavities for Second- and Third-Harmonic Generation

This paper presents an inverse-designed silicon nitride photonic crystal cavity that achieves a record-high quality factor of approximately 80,000 and demonstrates both second- and third-harmonic generation, establishing it as a superior platform for nonlinear and quantum photonics.

Original authors: M. Takiguchi, P. Heidt, X. Z. Lim, J. Zöllner, S. Yanagimoto, K. Nakayama, T. Aihara, M. Ono, H. Sumikura, M. Notomi

Published 2026-08-14

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: M. Takiguchi, P. Heidt, X. Z. Lim, J. Zöllner, S. Yanagimoto, K. Nakayama, T. Aihara, M. Ono, H. Sumikura, M. Notomi

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the world of light as a bustling city where information travels on beams of laser instead of cars on roads. In this city, engineers are constantly trying to build better "traffic control" systems to keep light moving fast, stay in one place, or change its color. To do this, they use tiny structures called photonic crystals. Think of these like a perfectly arranged forest of invisible trees (holes) that light has to navigate. If the trees are spaced just right, they can trap light in a specific spot, making it bounce around so much that it builds up incredible energy. This trapped light is called a cavity.

The goal for scientists is to make these cavities super efficient. They want a high Quality factor (Q), which is like how long a bell keeps ringing after you hit it (a high Q means it rings for a long time without losing energy), and a small Mode volume (V), which is how tiny the space is where the light is trapped. When you combine a long ring time with a tiny space, you get a massive amount of light squeezed into a microscopic area. This is the secret sauce for making light interact with matter, which is essential for things like ultra-fast computers, quantum computers, and sensors that can detect a single virus. However, building these perfect traps is tricky, especially when using a material called Silicon Nitride (SiN). While SiN is great because it's cheap, compatible with computer chips, and doesn't absorb light, it's not very good at holding onto light tightly because it's not "dense" enough optically. It's like trying to build a fortress out of light foam instead of solid stone; the light tends to leak out before it can build up enough power to do cool things.

This paper tells the story of how researchers at NTT and the University of Tokyo decided to stop trying to guess the perfect shape for their light fortress and instead let a computer figure it out. They used a technique called inverse design. Instead of starting with a shape and tweaking it a little bit, they told the computer, "We want the best possible light trap," and let the algorithm scramble the positions of the holes in the crystal until it found a solution that humans might never have thought of. They built a tiny, two-dimensional cavity made of Silicon Nitride and tested it. The results were impressive: they created a cavity that held light with a quality factor of about 80,000, which is the highest ever reported for this specific type of Silicon Nitride structure. But the real magic happened when they turned up the power. Because the light was trapped so tightly, it became intense enough to change its own color. They successfully observed the light doubling its frequency (Second-Harmonic Generation) and tripling it (Third-Harmonic Generation) right inside the same tiny box. This proved that their computer-designed shape was indeed squeezing the light into a tiny space with incredible strength, turning a material that usually just passes light through into a powerful engine for creating new colors of light.

Technical Summary: Inverse-Designed High-Q/V Silicon Nitride Photonic Crystal Cavities

Problem Statement
Silicon nitride (SiN) is a versatile platform for integrated photonics due to its CMOS compatibility, low optical loss, and broad transparency window from the visible to telecommunication wavelengths. Unlike silicon, SiN exhibits negligible two-photon absorption at telecommunication wavelengths, enabling high optical intensities for nonlinear applications. However, a significant limitation of SiN is its relatively low refractive index (approximately 2.0). This low index contrast with air weakens optical confinement, making it challenging to realize photonic crystal (PhC) cavities with simultaneously high quality factors ( $Q$ ) and small mode volumes ( $V$ ). Conventional deterministic optimization methods, which rely on small perturbations of air holes around a defect, have proven insufficient for achieving ultrahigh- $Q$ PhC cavities in stoichiometric SiN.

Methodology
To overcome the limitations of conventional design, the authors employed an inverse-design approach to optimize a two-dimensional SiN PhC cavity.

Design Strategy: The optimization targeted an $L6/5$ cavity geometry (a defect formed by removing holes and inserting additional holes, specifically an $L(x+1)/Lx$ structure). The initial geometry was chosen because different starting structures can converge to different local optima, and the $L6/5$ structure was expected to offer a superior $Q/V$ ratio compared to standard $L3$ or $L4/3$ designs.
Optimization Algorithm: The authors utilized a parallel simplex algorithm (PSA), a derivative-free optimization method. This approach was selected because it can be implemented with conventional electromagnetic simulators without requiring gradient information or large training datasets (unlike machine learning or automatic differentiation methods). The optimization parameters consisted of the positions of 54 air holes surrounding the cavity.
Simulation Parameters: The simulations assumed an air-bridge geometry with a slab thickness of 342 nm, a lattice constant of 686 nm, and a refractive index of 2.0. Mirror symmetry was imposed to reduce computational costs.
Fabrication: Devices were fabricated on a 425-nm-thick SiN film deposited via plasma-enhanced chemical vapor deposition (PECVD) on a silicon substrate. Patterns were defined by electron-beam lithography and etched using inductively coupled plasma (ICP). An air-bridge structure was formed by selectively undercutting the silicon substrate with potassium hydroxide (KOH) wet etching. The initial thickness was increased to 425 nm to compensate for thinning during the wet etching process, resulting in a final device thickness of approximately 350 nm.

Key Results

High-Q/V Ratio: The inverse-designed cavity achieved a simulated quality factor of $4.4 \times 10^5$ with a mode volume of approximately $0.96(\lambda/n)^3$ .
Experimental Demonstration: Experimental characterization of the fabricated devices yielded a quality factor of approximately 80,000 ( $8.0 \times 10^4$ ). The authors note this is the highest reported $Q$ factor for a near-stoichiometric SiN 2D PhC cavity. The experimental $Q$ factors showed excellent agreement with simulations across various waveguide coupling configurations.
Nonlinear Optical Generation: The strong optical confinement ( $Q/V$ $Q / V$ ) enabled the observation of both third-harmonic generation (THG) and second-harmonic generation (SHG) from the same cavity:
- THG: Observed when the fundamental cavity mode was resonantly excited using a femtosecond pulsed laser. THG was not observed when higher-order modes (with significantly lower $Q/V$ ) were excited, confirming that the effect relies on the high field enhancement of the fundamental mode. This demonstrates the utility of the cavity for intrinsic third-order nonlinear processes.
- SHG: Observed despite amorphous SiN being a centrosymmetric material with negligible bulk $\chi^{(2)}$ . The authors attribute this to an effective surface second-order susceptibility ( $\chi_s^{(2)}$ ) arising from symmetry breaking at the air–SiN interfaces. The strong localization of the optical field around the PhC air holes enhances the surface nonlinear response relative to the bulk, making SHG observable.
Robustness: The nonlinear effects were reproducibly observed across multiple fabricated devices, confirming that the performance stems from the optimized design rather than individual device anomalies.

Significance and Claims
The paper claims that this work establishes inverse-designed SiN PhC cavities as a promising platform for nonlinear and quantum photonics. The primary significance lies in demonstrating that inverse design can effectively overcome the optical confinement limitations inherent to low-refractive-index materials like SiN.

Key contributions highlighted by the authors include:

Overcoming Index Limitations: Proving that high- $Q/V$ cavities are achievable in stoichiometric SiN without resorting to silicon-rich compositions or nanobeam geometries, which are often more difficult to integrate.
Dual Nonlinear Capability: Providing experimental evidence that a single cavity can support both bulk third-order (THG) and surface-induced second-order (SHG) nonlinearities due to extreme field localization.
Platform Versatility: The demonstrated platform is compatible with a wide range of functional materials (e.g., GaN, ZnO, diamond, rare-earth ions, 2D materials), suggesting potential for heterogeneous integration in future photonic devices for wavelength conversion, enhanced light-matter interaction, and quantum photonics.

The authors remain modest regarding the specific applications, framing the results as a foundational step toward efficient nonlinear photonics and integrated devices rather than claiming immediate deployment in specific commercial systems. The work serves as a proof-of-concept that inverse design is a practical route to high-performance SiN nanocavities.

Technical Summary: Inverse-Designed High-Q/V Silicon Nitride Photonic Crystal Cavities

More like this