Inverse-designed release-free optomechanical crystal with high photon-phonon coupling

The authors present a release-free silicon optomechanical crystal that achieves a record vacuum optomechanical coupling rate of 800 kHz by combining human intuition with a novel multiphysics inverse-design algorithm, effectively bridging the performance gap between thermal robustness and strong photon-phonon coupling.

The Gist

Imagine a tiny, high-tech playground built inside a piece of silicon. On this playground, two invisible "dancers" perform: one is a beam of light (a photon), and the other is a vibration of the material itself (a phonon). The goal of this research is to get these two dancers to hold hands as tightly as possible so they can influence each other instantly. This interaction is the key to building future technologies that connect light-based internet signals with microwave-based quantum computers.

Here is the story of how the researchers solved a long-standing problem with this dance, explained simply:

The Problem: The "Floating" vs. "Anchored" Dilemma

In the past, scientists built these playgrounds by carving out tiny, floating beams of silicon (like a suspension bridge).

• The Good: Because they were floating, the light and vibration could get very close together, dancing tightly and efficiently.
• The Bad: Being floating made them fragile. When the light beam hit the silicon, it created heat. Since the beam was floating in the air, that heat had nowhere to go. It built up, causing the dance to get messy and noisy.

To fix the heat, scientists tried building "release-free" devices. These are like a dance floor that is glued firmly to the ground (the substrate).

• The Good: The heat drains away instantly into the ground, making the device very stable and quiet.
• The Bad: Because the floor was glued down, the vibration had to move in a very specific, fast way to stay trapped. This forced the light and vibration to dance in different spots, so they couldn't hold hands very tightly. The connection was weak.

The Trade-off: You could have a stable device with a weak connection, or a strong connection with a fragile, overheating device.

The Solution: A New Dance Move and a Smart Architect

The team at Chalmers University of Technology decided to break this rule. They wanted a device that was glued down (stable) but still had a super-strong connection. They did this in two steps:

1. The "X-HOPE" Trick (The Dance Move)
Imagine the playground is a hallway lined with mirrors (holes in the silicon). In previous designs, the light and vibration tried to meet in the middle, but the mirrors were spaced in a way that made the light spread out too much before it could grab the vibration.

The researchers used a clever trick called X-HOPE. They took pairs of mirrors and moved them closer together in a specific pattern.

• The Analogy: Think of it like a hallway where the walls suddenly pinch in. This forces the light beam to squeeze into a tiny, tight spot right in the center of the room.
• The Result: Because the light is now squeezed into a tiny spot, it lands exactly where the vibration is strongest. They are now dancing in the same spot, holding hands much tighter than before.

2. The Inverse-Design Algorithm (The Smart Architect)
Even with the new dance move, the playground wasn't perfect. The "mirrors" weren't reflecting the light and vibration perfectly, causing some energy to leak out.

Instead of trying to guess the perfect shape by hand, the researchers used a computer program called Inverse Design.

• The Analogy: Imagine you want a house with a specific view, perfect acoustics, and a specific number of windows. Instead of drawing a house and hoping it works, you tell a super-smart architect, "I want these results." The architect then works backward, erasing and rebuilding the walls millions of times in a split second until the house is perfect.
• The Result: The computer redesigned the shape of every single hole in the silicon, creating a complex, non-intuitive shape that traps the light and vibration perfectly, preventing any energy from leaking out.

The Result: A Record-Breaking Device

By combining the "X-HOPE" dance move with the "Smart Architect" design, they built a silicon chip that is:

• Glued down: It handles heat incredibly well, staying stable even under high power.
• Super-connected: The light and vibration interact with a strength (called the "coupling rate") of 800 kHz.

This is a record for devices that are glued down. It is now just as strong as the best "floating" devices ever made, but without the heat problems.

Why This Matters (According to the Paper)

The paper states that this achievement proves that "release-free" devices are now a viable platform for:

• Fast, low-noise classical computing: Processing signals quickly without errors.
• Quantum technologies: Helping to build systems that connect light to quantum computers (specifically mentioning "piezo-optomechanical microwave-to-optical transducers").

In short, they found a way to glue the dance floor down so it doesn't overheat, while using a clever trick and a super-computer to make the dancers hold hands tighter than ever before.

Technical Summary

Technical Summary: Inverse-designed release-free optomechanical crystal with high photon–phonon coupling

Problem Statement
Optomechanical crystals (OMCs) facilitate strong interactions between optical and mechanical fields, enabling applications in precision sensing, microwave-optical transduction, and quantum information processing. High-performance devices require both strong photon-phonon coupling rates and robustness against optical absorption-induced thermal noise. While suspended nanobeams offer tight mode confinement and high coupling, they suffer from poor thermal anchoring, making them susceptible to excess noise under optical pumping. Conversely, release-free OMCs, which remain mechanically attached to the substrate, provide superior thermal anchoring but have historically exhibited weaker vacuum optomechanical coupling rates ($g_{OM}$). This limitation arises because release-free operation typically necessitates a mechanical mode at the X-point and an optical mode near the half-X-point. In this configuration, the optical field extends gradually into the mirror region, resulting in sub-optimal spatial overlap with the tightly confined mechanical mode, thereby suppressing the coupling rate.

Methodology
The authors address this trade-off through a two-pronged approach combining a physics-guided design technique with a multiphysics inverse-design algorithm:

1. X-HOPE Design Technique: The authors introduce "X-point matching by harmonic period extension" (X-HOPE). By pairwise perturbing the unit cells of the crystal (specifically moving neighboring holes closer together), the periodicity is effectively doubled. This abruptly shifts the X-point of the band structure to match the wavevector of the optical mode ($k = 0.5\pi/a$). Consequently, a bandgap opens at the optical mode frequency, causing the optical field to decay immediately away from the defect. This concentrates the optical intensity at the center of the cavity, where the mechanical displacement is maximal, significantly improving spatial overlap compared to previous implementations where intensity peaked at the edges.

2. Multiphysics Inverse Design: While X-HOPE improves overlap, the resulting structures initially suffer from low radiation-limited quality factors ($Q$). To rectify this, the authors employ a gradient-based inverse-design algorithm using adjoint sensitivity analysis. This method optimizes over 200 variables (hole sizes and positions) in the transition region between the mirror and the defect. Unlike conventional adiabatic transitions, this approach handles the non-adiabatic transitions required by the X-HOPE geometry. The algorithm optimizes a figure-of-merit (FOM) based on the single-photon optomechanical scattering rate ($\Gamma_{OM}$), subject to constraints on mode frequencies, quality factors, and minimum feature sizes. To ensure robustness against fabrication imperfections, the geometry is stochastically eroded or dilated by up to 2 nm during each iteration.

Key Results
The authors fabricated and characterized the optimized silicon OMC on a SiO$_2$ substrate:

• Coupling Rate: The device achieved a record vacuum optomechanical coupling rate for a release-free architecture of $g_{OM}/(2\pi) \approx 800$ kHz (specifically 770–835 kHz depending on detuning). This is comparable to state-of-the-art suspended devices and represents a >50% improvement over previous release-free implementations.
• Scattering Rate: The resulting optomechanical scattering rate is $\Gamma_{OM}/(2\pi) = 1.1$ kHz, nearly double that of prior release-free devices.
• Quality Factors: The measured internal optical quality factor is $Q_o = 1.2 \times 10^5$, and the mechanical intrinsic linewidth is 11 MHz. The authors note that these values are lower than the simulated radiation-limited values ($Q_o > 10^6$, $Q_m = 20 \times 10^3$), attributing the discrepancy to sidewall scattering (due to etch angles) and Akhiezer damping at room temperature.
• Thermal Robustness: The device demonstrated resilience to optical heating, remaining unaffected at optical powers nearly 10 times higher than those causing redshifts in suspended devices.

Significance and Claims
The paper claims to have largely closed the performance gap between release-free and suspended optomechanical crystals. By combining the X-HOPE technique with multiphysics inverse design, the authors demonstrate that release-free architectures can achieve coupling strengths comparable to suspended devices while retaining their intrinsic thermal robustness. This establishes release-free OMCs as a viable platform for fast, low-noise classical and quantum optomechanics.

Furthermore, the authors posit that the inverse-design framework developed here is broadly applicable to the co-optimization of optical and mechanical resonances in other resonant systems. The X-HOPE technique is noted as relevant for any system requiring strong confinement or coupling for modes away from the Brillouin-zone edge, including multimode systems for harmonic generation and wave mixing. The work also provides a critical building block for the development of release-free piezo-optomechanical microwave-to-optical transducers.