🔬 optics

Giant backward Brillouin interaction in generic InP integrated photonics

This paper reports the first observation of giant backward stimulated Brillouin scattering in generic InP waveguides, achieving high gain coefficients of up to $737~\text{W}^{-1}\text{m}^{-1}$ mediated by weakly-guided pressure waves in a mature, foundry-accessible photonic platform.

Original authors: Riley te Morsche, Lukas Puts, Weiming Yao, David Marpaung

Published 2026-08-10

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Riley te Morsche, Lukas Puts, Weiming Yao, David Marpaung

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 internet as a massive, bustling highway where light beams act as cars, zooming through glass fibers to carry our videos, messages, and dreams. Usually, these light cars travel in perfect silence, but sometimes, they bump into the very road they're driving on. When a beam of light hits the tiny, invisible vibrations of the glass itself, it can bounce backward, losing a bit of energy and changing its color. This phenomenon is called "Stimulated Brillouin Scattering" (SBS). Think of it like a snowball rolling down a hill; if it hits a patch of rough ice, it might shatter and send a smaller snowball rolling back up the hill. For decades, scientists have tried to control this "backward snowball" effect because it can be a nuisance in long-distance cables, but also a superpower for building tiny, ultra-precise tools that process signals, isolate light, or even create lasers on a chip. The big challenge has been finding a material that is easy to build with, safe to handle, and powerful enough to make this effect happen reliably in the tiny circuits of modern electronics.

This paper reports a major step forward by catching this "backward snowball" effect in a material called Indium Phosphide (InP), which is already a workhorse in the photonics industry for making commercial optical transceivers. The researchers, working with a standard factory platform, successfully measured a very strong backward Brillouin interaction in these generic InP waveguides. They found that the light interacts with sound waves inside the chip to create a massive gain, measured at a coefficient of 737±54 W⁻¹m⁻¹. This is a huge number, placing InP on par with some of the best, but harder-to-use, materials like chalcogenides. The effect happens at a frequency shift of about 20.8 GHz and is incredibly narrow, with a linewidth of roughly 100 MHz.

To understand how this works, imagine the light wave as a heavy truck and the sound wave as a ripple in a pond. In these specific InP chips, the "pond" is shaped in a way that traps the ripples very tightly, even though the ripples are usually hard to contain. The researchers used computer simulations to figure out that the material's density and stiffness create a perfect environment for these ripples to stay confined, much like how a fiber optic cable traps light. The simulations showed that the light and sound waves overlap so perfectly that they boost each other's energy significantly. However, there was a small mystery: the computer models predicted the sound waves would vibrate at a slightly different speed than what the team actually measured. To fix this, the researchers adjusted their simulation by making the material slightly stiffer (by a factor of 1.21), which brought the numbers into perfect alignment with the real-world data. This suggests that while the basic physics is understood, there might be subtle properties in the real chips—perhaps related to electricity or tiny imperfections—that the simple models didn't fully capture yet.

The team measured this effect across several different chips and waveguides to be sure of their results. They found that the strength of the effect was consistent, with a normalized gain of 3.5±0.6 W⁻¹m⁻¹, which is comparable to other high-performance materials. They also noted that while their initial "free" simulations (without adjusting for stiffness) predicted a frequency shift of around 18.9 GHz, the actual measurement was 20.85 GHz. By tweaking the simulation to account for a stiffer acoustic environment, they could match the experiment almost exactly, though this adjustment lowered the predicted peak gain to 920 W⁻¹m⁻¹. The authors are careful to point out that they cannot yet confirm exactly why the real chips are stiffer than the standard models suggest; it could be due to piezoelectric effects (where electricity and movement mix) or free-carrier absorption, but these remain areas for future study.

Ultimately, this work proves that you don't need exotic, difficult-to-make materials to get powerful Brillouin effects. By using a mature, factory-ready Indium Phosphide platform, the researchers have opened the door to creating advanced signal processors, ultra-low-noise lasers, and sensitive sensors right on a single chip. Because this technology is already used to make the lasers and modulators that power our internet, adding this "backward scattering" superpower means we could soon see complex, all-in-one optical systems that are cheaper to build and easier to integrate with the electronics in our phones and computers. The paper suggests that this discovery could lead to a new generation of devices that process microwave and optical signals with incredible precision, all thanks to a better understanding of how light and sound dance together in a tiny, manufactured waveguide.

Technical Summary: Giant Backward Brillouin Interaction in Generic InP Integrated Photonics

Problem and Context
Stimulated Brillouin scattering (SBS) is a third-order nonlinear interaction between optical photons and GHz acoustic waves, widely utilized for applications such as microwave photonic signal processing, optical isolation, and distributed sensing. While established integrated platforms like chalcogenides (ChG), silicon-on-insulator (SOI), and silicon nitride (Si $_3$ N $_4$ ) have demonstrated SBS capabilities, they face specific limitations including material toxicity, complex fabrication requiring waveguide suspension, or relatively low gain coefficients. Transitioning these technologies to real-world environments is often hindered by propagation loss requirements and the inability to monolithically integrate active components like lasers and modulators. Indium phosphide (InP) is a mature, foundry-accessible platform capable of integrating these active components, yet the backward SBS interaction had not been previously measured or characterized in generic InP waveguides.

Methodology
The authors investigated backward, intramodal SBS in generic InP waveguides fabricated by Smart Photonics via the JePPIX multi-project wafer (MPW) service. The devices utilized shallow-etched rib waveguides (2 $\mu$ m wide) composed of In $_{1-x}$ Ga $_x$ As $_{1-y}$ P $_y$ /InP, with a core bandgap of 1.25 eV (corresponding to composition $y=0.53$ ).

The experimental approach involved:

Measurement: A pump-probe measurement scheme using intensity modulation and lock-in detection to measure the SBS gain spectrum. The waveguide gain coefficient was calibrated against a fiber response measured in the same setup.
Statistical Analysis: Measurements were repeated across multiple waveguide samples on two chip samples to determine experimental uncertainty and sample variations.
Numerical Simulation: Theoretical modeling was conducted using a formalism described in prior literature [45]. Simulations calculated the TE-like optical modes and acoustic responses. To reconcile discrepancies between simulation and experiment, the authors performed "fit-free" simulations and subsequently adjusted the acoustic stiffness by a factor of 1.21 to emulate effective mode stiffness changes.

Key Results
The study reports the first measurement of backward SBS in generic InP waveguides, yielding the following key findings:

High Gain Coefficients: The experiments observed SBS gain coefficients as high as $g_B = 737 \pm 54$ W $^{-1}$ m $^{-1}$ at a frequency shift of $\Delta\nu_B \approx 20.85$ GHz.
Normalized Gain: The normalized Brillouin gain coefficient, mediated by weakly-guided pressure waves in InGaAsP, was determined to be $g_B/Q_m = 3.5 \pm 0.6$ W $^{-1}$ m $^{-1}$ . This value is comparable to As $_2$ S $_3$ (ChG) and significantly outperforms other scalable platforms like Si $_3$ N $_4$ .
Linewidth: The measured Brillouin linewidth was approximately $\Gamma_B \approx 104$ MHz.
Simulation vs. Experiment: Fit-free simulations predicted a frequency shift of $\approx 18.9$ GHz and a gain of $1110 $W$ ^{-1} $m$ ^{-1}$. To match the experimental frequency shift of 20.8 GHz, a simulation with 1.21x increased acoustic stiffness was required, which lowered the predicted peak gain to $920 $W$ ^{-1} $m$ ^{-1}$. The authors note that the discrepancy suggests potential differences between assumed material properties and the physical sample, possibly due to overlooked physical processes such as piezoelectricity or carrier-induced modulation, though these mechanisms were not confirmed.

Significance and Claims
The paper claims that this work expands the nonlinear toolbox of the mature, foundry-accessible InP platform by adding backward SBS. The authors position InP as a competitive platform for Brillouin photonics, noting that its gain coefficient places it on par with chalcogenides while offering the advantage of monolithic integration with active components (lasers, modulators, detectors) and electronic interfacing.

The authors suggest that the narrow linewidth ( $\approx 100$ MHz) and high gain observed in this foundry platform open new opportunities for high-performance microwave and optical signal processing. Specifically, they identify potential for:

Low-noise Brillouin lasers, potentially operating in the phonon lasing regime.
Advanced microwave photonics applications that can circumvent strict propagation loss requirements due to the high gain.
The development of monolithic Brillouin photonics systems leveraging existing foundry building blocks.

The study concludes that while the current simulations provide a basic understanding of the mechanism, the observed frequency shifts and gain values indicate that the physical sample properties may differ from standard literature values, warranting further investigation into the specific acousto-optic and piezoelectric properties of the InP waveguides.

More like this