Stable, bidirectional electro-optic transduction in thin film lithium tantalate

This paper demonstrates the first stable, bidirectional electro-optic microwave-to-optical transduction in thin-film lithium tantalate (TFLT), achieving high-efficiency coherent conversion with minimal added noise and superior long-term bias stability to enable scalable quantum interconnects.

The Gist

Imagine you have two very different languages that need to talk to each other. One language is spoken by superconducting quantum computers (which use microwave signals, like the Wi-Fi in your home but much faster and more delicate). The other language is spoken by fiber optic cables (which use light, or photons, to send information across the world).

Right now, these two languages can't understand each other. To build a "quantum internet" that connects many quantum computers together, we need a translator. This paper introduces a new, highly effective translator made from a special material called Thin-Film Lithium Tantalate (TFLT).

Here is a breakdown of what the researchers achieved, using simple analogies:

1. The Problem with the Old Translators

Previously, scientists tried to build these translators using a material called Lithium Niobate. It worked okay, but it had a major flaw: it was like a radio that constantly drifts out of tune. To keep it working, you had to constantly adjust the volume knob (apply a "bias voltage") to stop the signal from fading. This made the devices complicated and hard to scale up for mass production.

2. The New Solution: A "Stable" Material

The team switched to Lithium Tantalate. Think of this material as a tuning fork that never loses its pitch.

• The Analogy: If the old material was a rubber band that stretched and needed constant re-tightening, the new material is a solid steel rod.
• The Result: They built a translator that stays perfectly tuned for days at a time without needing any constant adjustments. They just set it once, and it works.

3. How the Translator Works (The "Photonic Molecule")

Inside the chip, the researchers built a tiny machine with three main parts:

• Two Optical Resonators: Imagine two race tracks for light particles (photons) running side-by-side. They are so close that the light can "leak" from one track to the other, creating a synchronized dance called a "photonic molecule."
• One Microwave Resonator: This is a superconducting loop that catches microwave signals.
• The Interaction: When you shine a laser (the pump) into the system, it acts like a conductor. It takes a microwave signal (the input) and converts it into a light signal (the output), or vice versa.

The magic happens because the two light tracks are tuned to specific frequencies that match the microwave signal, allowing the energy to swap back and forth efficiently.

4. Mass Production: From Hand-Crafted to Factory-Made

Most previous quantum devices were made using a technique called "electron-beam lithography," which is like drawing each device by hand with a super-fine pen. It's slow and you can only make a few at a time.

This team used Deep Ultraviolet Lithography (DUVL), which is like using a stencil and a spray paint gun to print hundreds of devices on a single silicon wafer at once.

• The Result: They successfully made hundreds of these translators on one chip, and they all worked almost exactly the same way. This proves the technology can be scaled up for real-world use.

5. Performance and Stability

• Efficiency: The translator is good at its job. It successfully converted signals back and forth between light and microwaves with a coupling rate (how fast they talk) of about 1,000 times per second per photon. This matches what the math predicted.
• Noise: When you translate, you sometimes introduce "static" (noise). The team found that by using short pulses of light (like a camera flash) instead of a continuous beam, they could keep the noise incredibly low—less than one extra "grain of static" (photon) for every 100 microseconds of operation.
• Longevity: They ran the device continuously for several days. Because the material is so stable, they didn't need to fiddle with the settings, proving it's ready for long-term use.

Summary

In short, this paper presents a new, stable, and mass-producible translator that allows quantum computers (which speak microwaves) to talk to the internet (which speaks light). By using a material that doesn't drift out of tune and a manufacturing method that allows for mass production, the researchers have taken a significant step toward building a future where quantum computers can be linked together over long distances.

Technical Summary

Technical Summary: Stable, Bidirectional Electro-Optic Transduction in Thin Film Lithium Tantalate

Problem Statement
Efficient and stable microwave-to-optical transduction is a critical enabling technology for distributed superconducting quantum computing and heterogeneous quantum networks. While electro-optic (EO) transducers based on thin-film lithium niobate (TFLN) have shown promise, their practical deployment is currently hindered by several factors: low-frequency bias drift requiring active compensation, limited efficiency, fabrication complexity, and scalability issues. Furthermore, indirect transduction approaches (e.g., piezo-opto-mechanical) often suffer from bandwidth limitations and high added noise due to intermediate modes. Direct EO approaches using the Pockels effect offer a pathway to high efficiency but require materials that combine strong nonlinearity with intrinsic stability and high-power handling capabilities.

Methodology
The authors present the first realization of integrated electro-optic microwave-optical transducers using thin-film lithium tantalate (TFLT). The device architecture employs a triply resonant design consisting of:

1. Optical System: Two evanescently coupled lithium tantalate rib waveguide racetrack resonators forming a "photonic molecule." This supports symmetric (S) and antisymmetric (AS) supermodes. The frequency difference between these supermodes ($\omega_S - \omega_{AS}$) is engineered to match the microwave resonator frequency ($\omega_m$).
2. Microwave System: A quasi-lumped-element LC superconducting resonator coupled capacitively to a coplanar waveguide (CPW).
3. Tuning Mechanism: Dedicated DC electrodes tune the phase of the "dark" resonator to match the "bright" resonator, creating the necessary avoided crossing for transduction.
4. Fabrication: The devices are fabricated using wafer-scale deep ultraviolet lithography (DUVL) on a commercially available stack of 400 nm X-cut lithium tantalate on SiO$_2$/Si. The process involves patterning optical waveguides and a superconducting NbN film via lift-off to avoid etch damage. Superconducting air bridges are used to route electrical signals over optical layers without lossy cladding materials.

Key Contributions

• Material Platform: The demonstration of TFLT as a viable alternative to TFLN, leveraging lithium tantalate's comparable Pockels nonlinearity but superior low-frequency stability and resistance to photorefractive effects.
• Scalable Fabrication: The use of DUVL to produce hundreds of devices per wafer, overcoming the throughput limitations of electron-beam lithography typically used for integrated EO converters.
• Bidirectional Operation: Successful demonstration of coherent transduction in both directions (microwave-to-optical and optical-to-microwave) across six distinct devices.
• Stability: Achievement of continuous operation over multiple days using a single static bias field with minimal feedback, eliminating the need for active bias drift compensation.

Results

• Efficiency and Coupling: The team observed coherent bidirectional conversion between C-band optical photons and 4.9–5.5 GHz microwave photons. Measured on-chip efficiencies aligned with theoretical predictions, yielding inferred single-photon coupling rates of $g_0/(2\pi) \sim 1$ kHz (specifically up to 1.1 kHz in Devices 1 and 4).
• Optical Loss: Wafer-scale measurements revealed tightly clustered resonator losses, with post-process transducer modes centered near 229 MHz internal loss rates. Equivalent propagation losses in TFLT were measured as low as 0.25 dB/cm.
• Stability: The devices operated stably for hours to days under continuous wave (CW) and pulsed pumping conditions. Unlike TFLN devices which often require swept bias voltages, these TFLT devices maintained performance with a static bias.
• Added Noise: Under pulsed pumping, the transducers added less than one photon of noise for 100 $\mu$s pulses at the highest measured efficiencies. Noise characterization indicated that scattered light from grating couplers was the dominant source of added noise, rather than on-chip waveguide scattering.
• Cryogenic Behavior: The devices showed a reduction in the apparent electro-optic coefficient ($r_{33}$) when cooled from room temperature to <4 K, a phenomenon consistent with reports on other EO materials. Despite this, the extracted $g_0$ values remained high.

Significance
The paper establishes TFLT as a scalable, robust, and stable platform for future quantum interconnects. The intrinsic DC bias stability of lithium tantalate addresses a major operational hurdle in scaling quantum transducers, allowing for simplified device control in multi-channel architectures. By combining wafer-scale fabrication with high-performance superconducting integration, this work demonstrates a pathway toward the large-scale production of optical interconnects necessary for modular quantum processors. The authors posit that with further optimization of optical losses, microwave quality factors, and coupling strengths, TFLT transducers could enable high-fidelity entanglement generation between superconducting qubits over optical fiber networks with minimal protocol overhead.