🔬 applied physics

Free-Space Characterization Setup for Low-loss Aluminum Oxide Waveguides at 261 nm

This paper presents a free-space imaging methodology for characterizing deep-ultraviolet polycrystalline aluminum oxide waveguides at 261 nm, demonstrating a low-loss propagation of approximately 4.6 dB/cm for 600 nm wide structures while establishing a foundational analysis pipeline for future deep-UV photonic integrated circuit validation.

Original authors: Vahram Voskerchyan (Bonnie), Dawson Bonneville (Bonnie), N. X. Xu (Bonnie), Michiel de Goede, Ward Henrdiks, Lantian Chang, S. M. Garcia Blanco

Published 2026-03-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Vahram Voskerchyan (Bonnie), Dawson Bonneville (Bonnie), N. X. Xu (Bonnie), Michiel de Goede, Ward Henrdiks, Lantian Chang, S. M. Garcia Blanco

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 you are trying to build a super-fast, microscopic highway for light, but instead of driving cars, you are guiding beams of ultraviolet (UV) light. This isn't just any light; it's the kind used for advanced medical tools, quantum computers, and high-precision measurements. The problem? Most existing "roads" (materials) melt or absorb this light before it can get anywhere, making it impossible to build efficient circuits.

This paper is like a blueprint and a field report from a team of engineers who successfully built a new type of highway using Aluminum Oxide (the same stuff in sapphire jewelry) and figured out exactly how well it works.

Here is the story of their experiment, broken down into simple concepts:

1. The Challenge: Finding the Right Material

Think of the ultraviolet spectrum (200–400 nm) as a "dangerous zone" for light.

Silicon Nitride (a common material for light circuits) is like a sponge here; it soaks up the UV light immediately, causing massive signal loss.
Aluminum Oxide (Al₂O₃), however, is like a clear glass tunnel. It lets this harsh UV light pass through without eating it up. The team wanted to prove they could make a "road" out of this material that was smooth enough for light to travel long distances without getting lost.

2. Building the Road (The Setup)

The team didn't just pour concrete; they used high-tech construction:

The Foundation: They started with a silicon wafer (like a giant cookie sheet) and sprayed a very thin layer of Aluminum Oxide onto it, about as thick as a strand of spider silk.
The Pattern: Using a super-precise "electron pen" (Electron-Beam Lithography), they drew tiny lines on this layer.
The Tunnel: They etched these lines to create channels (waveguides) where the light would travel.
The Shield: They covered the whole thing with a protective glass layer (Silicon Dioxide) to keep the light inside the tunnel, like putting a lid on a pipe.

3. The Test Drive: How to Measure the Loss

This is the most clever part of the paper. Usually, to measure how much light is lost on a road, you need to hook up a sensor at the start and another at the very end. But at these tiny scales and with this dangerous UV light, that's hard to do.

Their Solution: The "Leaky Pipe" Analogy
Imagine a garden hose with tiny pinholes in it. As water flows through, a little bit sprays out of the holes. If you look at the hose from the side, you can see a line of mist.

The Mist: The UV light traveling through the waveguide isn't perfectly contained; a tiny bit "leaks" out the sides as scattered light.
The Camera: The team set up a special UV-sensitive camera to take a picture of this "mist" (scattered light) along the entire length of the waveguide.
The Math: They looked at the photo and asked: "How much dimmer is the mist at the end of the road compared to the start?"
- If the mist is bright at the start and barely visible at the end, the road is "leaky" (high loss).
- If the mist stays bright all the way down, the road is smooth (low loss).

They used a mathematical formula (exponential decay) to calculate exactly how much energy was disappearing per centimeter, just by analyzing the brightness of the scattered light in the photo.

4. The Results: Wide Roads vs. Narrow Roads

They tested roads of different widths, from very narrow (200 nanometers) to wide (2000 nanometers).

The Narrow Roads (200 nm): These were like driving a semi-truck through a narrow alleyway. The light bounced off the rough walls (sidewall scattering) and got lost easily. The loss was high (about 18 dB/cm).
The Wide Roads (600 nm+): These were like a wide highway. The light had plenty of room to stay in the middle, away from the rough walls. The loss dropped significantly to about 4.6 dB/cm.

Why does this matter?
In the world of UV light, losing 4.6 dB per centimeter is actually a huge victory. It means the light can travel far enough to be useful for real-world devices.

5. The Safety First Approach

Because UV light at this wavelength (261 nm) is like a super-powerful sunburn ray that can damage eyes and skin instantly, the entire experiment was built inside a fortress.

The setup was completely enclosed.
They used special filters to block any stray, dangerous light.
It was a "safe zone" for the researchers to play with dangerous light.

The Big Picture Takeaway

This paper is essentially a "How-To" guide for the future. The authors aren't just saying, "We made a good road." They are saying, "Here is exactly how we built the testing equipment, how we took the pictures, and how we calculated the results so you can do it too."

They proved that Aluminum Oxide is a promising material for building tiny, efficient circuits that work with deep UV light. This opens the door for smaller, cheaper, and more powerful devices in fields like medical diagnostics, quantum computing, and advanced microscopy.

In a nutshell: They built a smooth, clear tunnel for dangerous UV light, used a camera to watch the light "leak" out the sides to measure how well it traveled, and found that wider tunnels work much better than narrow ones. Now, other scientists can use their method to build even better devices.

1. Problem Statement

The ultraviolet (UV) spectral range (200–400 nm) is critical for applications such as UV Raman spectroscopy, microscopy, metrology, and quantum computing. However, integrating photonic circuits in this range is challenging due to material limitations:

Silicon Nitride (SiN): While mature, it suffers from prohibitively high absorption losses below 450 nm due to its small bandgap.
III-Nitrides (AlN/AlGaN): These have large bandgaps suitable for deep UV, but current implementations suffer from high propagation losses, limiting their utility.
Silicon Oxynitride (SiOxNy): Offers low loss but lacks the high refractive index contrast required for high-density integration.

Aluminum Oxide (Al2O3) is identified as a promising alternative due to its large transparency window (UV to mid-IR) and compatibility with CMOS processes. However, characterizing low-loss propagation in the deep UV (specifically around 261 nm) for polycrystalline Al2O3 waveguides remains a technical hurdle requiring specialized measurement techniques that ensure operator safety and data accuracy.

2. Methodology

The authors developed a comprehensive experimental setup and analysis pipeline to characterize waveguide attenuation at 261 nm.

A. Device Fabrication

Material Deposition: Polycrystalline Al2O3 layers (~95 nm thick) were deposited on 100 mm silicon wafers (with 8 µm thermal oxide) using RF reactive co-sputtering (AJA ATC 15000).
Patterning & Etching: Waveguides were defined via Electron-Beam Lithography (EBL) and etched using Reactive Ion Etching (RIE).
Cladding & Annealing: The structures were clad with 8 µm SiO2 (deposited via LPCVD and PECVD) and annealed to improve film quality.
Dimensions: Waveguide widths tested ranged from 200 nm to 2 µm, with a minimum bend radius of 450 µm.

B. Experimental Setup (Free-Space Coupling)

Light Source: A 261 nm UV laser coupled into the waveguide via a high-NA objective lens (Thorlabs LMU-20X-UVB).
Safety & Filtering: The entire setup was enclosed in a protective housing with UV-blocking filters to prevent skin/eye damage. A UG11-UV filter removed residual shorter wavelengths, ensuring only 261–300 nm light was present.
Detection: A UV-sensitive monochrome camera captured the scattered light intensity along the waveguide. A spectrometer at the output facet confirmed the spectral purity (261 nm only).
Alignment: A 5-axis piezoelectric stage ensured precise alignment, achieving a mode-overlap efficiency of ~46%.

C. Image-Based Analysis Pipeline

To extract propagation losses, the authors employed a rigorous image processing workflow:

Spatial Calibration: Converted camera pixel coordinates to physical distances on the chip using lens magnification and pixel pitch.
Data Segmentation: The waveguide image was divided into spatial blocks to average intensity, reducing sensitivity to localized scattering or pixel noise.
Linear Range Verification: Exposure parameters were tuned to ensure the camera operated within its linear dynamic range; saturated pixels were excluded.
Exponential Fitting: The scattered intensity $I(x)$ was fitted to the model $I(x) = I_0 \exp(-\alpha x)$ to determine the attenuation coefficient $\alpha$ .
Loss Calculation: Propagation loss was calculated as $10 \log_{10}(I_{max}/I_{min})$ .

3. Key Results

The study successfully quantified propagation losses for various waveguide widths at 261 nm:

600 nm Waveguide: Achieved a propagation loss of 4.6 dB/cm.
200 nm Waveguide: Exhibited significantly higher losses of approximately 18 dB/cm.
Trend: Losses decreased as waveguide width increased (200 nm → 2000 nm).
- Narrow Waveguides (200 nm): Suffered from strong sidewall scattering and poor mode confinement. The small width increased the overlap of the optical mode with etched sidewalls, where roughness causes significant Rayleigh scattering (scaling as $1/\lambda^4$ ).
- Wider Waveguides (600 nm+): Showed improved optical confinement and reduced interaction with sidewall roughness, leading to lower attenuation.
Bend Loss: The 450 µm bend radius was sufficient for wider waveguides but contributed to additional radiative leakage in the narrowest (200 nm) structures due to reduced mode confinement.

4. Key Contributions

Methodological Framework: The paper documents a robust, free-space, image-based characterization methodology specifically tailored for deep-UV photonic integrated circuits (PICs).
Safety & Precision: Demonstrates a fully enclosed, filtered experimental setup that balances operator safety (against UVC hazards) with high-precision optical measurement.
Material Validation: Validates RF reactive sputtered polycrystalline Al2O3 as a viable platform for deep-UV applications, achieving losses comparable to other state-of-the-art platforms (e.g., ~4.6 dB/cm at 261 nm).
Scalability Insight: Provides critical data on the relationship between waveguide geometry and loss in the deep UV, highlighting the trade-offs between integration density (narrow waveguides) and propagation loss.

5. Significance

This work addresses a critical gap in the development of deep-UV photonic technologies. By establishing a reliable characterization pipeline and demonstrating low-loss propagation in Al2O3, the authors provide a foundation for:

Scalable UV-PICs: Enabling the development of compact, robust, and cost-effective devices for UV spectroscopy and quantum computing.
Process Optimization: Offering a benchmark for future fabrication improvements, particularly regarding sidewall roughness control in deep-UV regimes.
Future Research: The methodology serves as a template for further validation and refinement of loss measurements, with the authors noting that independent techniques are being investigated to further verify these initial results.

In summary, this paper successfully bridges the gap between material deposition and functional device characterization in the deep ultraviolet, proving that Al2O3 is a competitive platform for next-generation UV integrated photonics.