🔬 optics

Low voltage and high-bandwidth thin-film lithium tantalate modulator on a silicon dioxide substrate

This paper reports the first fabrication and characterization of a low-voltage, high-bandwidth thin-film lithium tantalate electro-optic modulator on a fused-silica substrate, which achieves a 64 GHz bandwidth, a 1.53 V half-wave voltage, and a net 440.6 Gbps data rate, demonstrating its potential as a superior alternative to lithium niobate for next-generation integrated photonics.

Original authors: Zihan Li, Alexander Kotz, Adrian Schwarzenberger, Christian Koos, Tobias J. Kippenberg

Published 2026-04-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Zihan Li, Alexander Kotz, Adrian Schwarzenberger, Christian Koos, Tobias J. Kippenberg

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, high-speed highway system. Right now, our data centers (the giant warehouses where all the internet's traffic is processed) are growing so fast that the roads are getting clogged. The old "electrical wires" used to move data are like narrow, pothole-ridden dirt paths; they can't handle the sheer volume of traffic or the speed required for modern Artificial Intelligence.

To fix this, engineers are building "optical highways"—using light instead of electricity to carry data. But to get data onto these light highways, you need a traffic controller called a modulator. This device turns electrical signals (your data) into flashes of light.

For a long time, the best traffic controllers were made from a material called Lithium Niobate. They are great, but they are expensive to make and have a few annoying quirks, like getting "drifty" (losing their calibration over time) and struggling with high-power light.

Enter the star of this paper: Lithium Tantalate.

The Problem: The "Heavy" Substrate

Think of a modulator like a race car. The car body is the Lithium Tantalate, but it needs a chassis (a substrate) to sit on.

The Old Way: Previously, scientists built these Lithium Tantalate cars on a Silicon chassis. Silicon is like a heavy, dense concrete block. While it's cheap, it messes with the car's suspension. It creates too much "friction" (electrical loss) for the microwave signals that drive the modulator, limiting how fast the car can go.
The New Way: This paper introduces a Fused Silica (basically, super-pure glass) chassis. Think of this as switching from a heavy concrete block to a sleek, lightweight carbon-fiber frame. It's much smoother and lets the signals travel with almost no friction.

The Innovation: The "Slow-Wave" Superhighway

Even with a better chassis, you still need to make sure the "driver" (the microwave signal) and the "passenger" (the light signal) are moving at the exact same speed. If the driver is too fast or too slow, the data gets garbled.

The researchers used a clever trick called a Slow-Wave Electrode.

The Analogy: Imagine a runner trying to keep pace with a cheetah. If the runner runs on flat ground, they can't keep up. But if the runner runs on a track with speed bumps (the segmented T-shaped electrodes), they are forced to slow down.
The Result: By adding these "speed bumps," the microwave signal is slowed down just enough to perfectly match the speed of the light. This "velocity matching" allows the system to handle incredibly high speeds without losing synchronization.

The Results: Breaking the Speed Limit

By combining the Glass Chassis (Fused Silica) with the Speed-Bump Track (Slow-Wave Electrodes), the team built a modulator that is:

Low Voltage: It takes very little energy to switch the light on and off (like a gentle tap on the brake instead of a slam).
Super Fast: It can handle data speeds up to 64 GHz right now, with the potential to hit 100 GHz.
Stable: Unlike its Lithium Niobate cousin, it doesn't get "drifty." It stays perfectly calibrated even after running for a long time.

The Real-World Test: The Data Marathon

To prove it works, they didn't just measure speed; they sent actual data through it.

They used a coding scheme called PAM8 (which is like sending messages using 8 different shades of gray instead of just black and white).
The Score: They achieved a net data rate of 440.6 Gigabits per second.
What does that mean? That's fast enough to download a high-definition movie in less than a second, or stream thousands of 4K videos simultaneously on a single fiber.

Why This Matters

This isn't just a lab experiment. Lithium Tantalate is already used in millions of cell phone filters (the things that stop your phone from picking up the wrong radio stations). Because the manufacturing process for this material is already mature and cheap, switching to this new "Glass + Speed-Bump" design could make next-generation internet hardware cheaper, faster, and more energy-efficient than what we have today.

In short: The researchers found a way to build a better, faster, and cheaper traffic controller for the internet's light highways, solving the speed limits that were holding us back.

1. Problem Statement

The rapid growth of artificial intelligence and data centers has created an urgent demand for higher transmission bandwidths and energy efficiency in optical interconnects. While Thin-Film Lithium Niobate (TFLN) modulators are currently the state-of-the-art, they face challenges including high fabrication costs, DC bias instability, and photorefractive effects under high optical power.

Lithium Tantalate (LiTaO₃, LT) is a promising alternative material offering lower birefringence, reduced DC drift, and higher optical power handling. However, previous LT-based photonic integrated circuits (PICs) were fabricated on silicon substrates. The high dielectric constant of silicon ( $\epsilon_r \approx 11.7$ ) creates significant challenges for traveling-wave electrode design:

It limits the ability to achieve velocity matching between microwave and optical signals without excessive capacitance.
It forces narrow electrode gaps to maintain impedance, leading to high conductor losses.
These factors previously limited LT modulators to short interaction lengths (~6 mm), resulting in high half-wave voltages ( $V_\pi \approx 4.8$ V) and restricted bandwidths.

2. Methodology

The authors propose a novel platform: Thin-Film Lithium Tantalate on Fused Silica (LT-on-FS). Fused silica has a low dielectric constant, enabling better microwave propagation and electrode design.

Key Methodological Steps:

Substrate Transfer: An X-cut thin-film lithium tantalate wafer (600 nm LiTaO₃ on a 2 µm SiO₂ buffer and Si handle) was transferred to a 4-inch (100 mm) fused silica wafer via direct wafer bonding using oxygen plasma activation. The original silicon carrier and buffer were removed via grinding and etching, achieving a >95% transfer yield.
Device Fabrication:
- Waveguides: Defined using Deep-UV (DUV) lithography and ion beam etching, leaving a 160 nm slab for efficient modulation.
- Electrodes: A segmented T-shaped slow-wave Coplanar Waveguide (CPW) was designed. This design uses periodic capacitive loading to separate current paths, allowing for a wider center conductor (reducing loss) while maintaining 50 $\Omega$ impedance and matching the microwave phase velocity to the optical group velocity.
- Singulation: Chips were separated using femtosecond laser stealth dicing to avoid the issues associated with dicing amorphous fused silica.
Characterization:
- Static: Measured $V_\pi$ and extinction ratios at 1300 nm (O-band) and 1550 nm (C-band).
- Dynamic: Used a Vector Network Analyzer (VNA) up to 125 GHz to measure S-parameters ( $S_{11}, S_{21}$ ) and electro-optic (EO) frequency response.
- System-Level: Conducted high-speed Intensity Modulation and Direct Detection (IMDD) experiments using Pulse-Amplitude Modulation (PAM4, PAM6, PAM8) to test net data rates.

3. Key Contributions

First LT-on-FS Modulator: This is the first demonstration of thin-film lithium tantalate modulators fabricated on a fused silica substrate, moving away from silicon to overcome dielectric limitations.
Optimized Slow-Wave Electrode: Implementation of a segmented T-shaped electrode design that successfully balances velocity matching, impedance matching, and microwave loss reduction on a low-permittivity substrate.
Scalability: Demonstration of wafer-scale fabrication on a 4-inch fused silica wafer with uniform performance across the chip.
Performance Benchmarking: Achieved performance metrics comparable to state-of-the-art TFLN modulators, validating LT as a viable alternative for next-generation links.

4. Key Results

Low Voltage & High Efficiency:
- Achieved a half-wave voltage ( $V_\pi$ ) of 1.53 V at 1550 nm and 1.21 V at 1300 nm for an 18 mm-long device.
- The $V_\pi L$ product was approximately 2.86 V·cm (C-band) and 2.42 V·cm (O-band).
High Bandwidth:
- Demonstrated a 3-dB electro-optic bandwidth of 64 GHz.
- The 6-dB bandwidth exceeded 100 GHz.
- Note: The roll-off above 60 GHz was attributed to an inadvertent optical path imbalance (used for spectral biasing) rather than fundamental microwave loss. Simulations suggest that with an optimized design, a 100 GHz 3-dB bandwidth is achievable with the same $V_\pi$ .
Low Loss & Stability:
- Microwave attenuation was measured at approximately 4.6 dB/cm at 120 GHz.
- Exhibited low bias drift with stable switching voltage down to 10 mHz.
High-Speed Data Transmission:
- Achieved a net data rate of 440.6 Gbit/s using PAM8 signaling at a symbol rate of 176 GBd.
- This performance is comparable to the best-reported TFLN modulators on quartz substrates.

5. Significance

This work establishes Thin-Film Lithium Tantalate on Fused Silica as a scalable, high-performance alternative to Lithium Niobate for electro-optic links.

Material Advantages: LT offers inherent advantages over LN, including lower birefringence (simplifying complex PIC designs like AWGs), better DC stability, and higher power handling.
Manufacturing Viability: By utilizing fused silica and leveraging mature foundry infrastructure for LT (already used in RF filters), the platform promises cost-effective, wafer-scale manufacturing comparable to Silicon-on-Insulator (SOI) but with superior electro-optic performance.
Future Impact: The results pave the way for energy-efficient, high-bandwidth optical interconnects essential for future AI clusters and data centers, potentially replacing or complementing current TFLN solutions.