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

Abstract: Efficient and stable microwave-optical transduction is a key enabling technology for distributed superconducting quantum computing and heterogeneous quantum networks. Electro-optic transducers based on thin-film lithium niobate (TFLN) have shown strong promise, but demonstrations to date have been limited by various factors such as low frequency bias drift, low efficiency, fabrication complexity, and scalability. Here we demonstrate the first integrated electro-optic microwave-optical transducers realized in thin-film lithium tantalate (TFLT), a material platform offering Pockels nonlinearity comparable to TFLN together with improved bias stability and high-power handling. We fabricate superconducting microwave resonators coupled to tunable photonic-molecule optical resonators using wafer-scale deep ultraviolet lithography, offering high-throughput production of hundreds of devices per wafer. Across six devices we observe coherent bidirectional conversion between C-band optical photons and 4.9-5.5 GHz microwave photons, with measured on-chip efficiencies and inferred single-photon coupling rates g_0/2π ~ 1 kHz consistent with theory. Continuous operation over multiple days is achieved using a static bias field with minimal feedback, demonstrating a major operational advantage. We further characterize optical loss statistics, microwave resonator performance, and optically induced added noise under pulsed pumping, finding less than one added photon for 100 microsecond pulses at the highest measured efficiencies. These results establish TFLT as a scalable and robust electro-optic platform for future quantum interconnects and modular quantum processors.

• The paper demonstrates a transducer design achieving stable, bidirectional conversion between C-band optical photons and 4.9–5.5 GHz microwave photons with efficiency tracking predicted single-photon EO coupling rates of around 1 kHz.
• It employs a triply resonant architecture with optical supermodes and a superconducting microwave LC resonator, leveraging wafer-scale deep ultraviolet lithography for high device yield and uniform performance.
• The work shows significant operational benefits over TFLN by offering stable DC bias, reduced photorefractive effects, and potential for enhanced quantum interconnects via process optimization.

Integrated Electro-Optic Microwave–Optical Transduction in Thin-Film Lithium Tantalate

Introduction

The demonstrated integration of stable, bidirectional electro-optic (EO) transduction between microwave and optical frequencies in thin-film lithium tantalate (TFLT) devices constitutes a significant milestone for scalable quantum interconnects. The work addresses critical limitations observed in thin-film lithium niobate (TFLN)-based converters, such as DC bias drift, fabrication complexity, and insufficient scalability. Here, the authors leverage the advantageous material properties of TFLT—wider bandgap, reduced photorefractive susceptibility, higher power-handling, and superior DC bias stability—to develop an integrated transducer platform supporting high-throughput fabrication.

Device Architecture and Fabrication

The TFLT transducers are implemented via a triply resonant architecture, employing two coupled rib waveguide optical racetrack resonators forming a photonic molecule, alongside a superconducting quasi-lumped LC microwave resonator. Electrodes are patterned by wafer-scale deep ultraviolet lithography (DUVL), yielding hundreds of devices per wafer with low device-to-device variance, directly addressing production scalability. Waveguides are air-clad, avoiding oxide claddings with deleterious microwave loss tangents and eliminating piezoelectric parasitic modes by aggressive etching of extraneous TFLT.

Phase and frequency matching of the optical supermodes (symmetric S and antisymmetric AS) and the microwave mode is tunable via DC electrodes, enabling reliable hybridization for both microwave–optical and optical–microwave transduction. The robust performance of this architecture is supported by precise control over coupling rates, electrode coverage, geometric design parameters, and process uniformity.

Experimental Assessment of Transduction

Efficiency and Conversion

The devices exhibit coherent bidirectional conversion between C-band optical and 4.9–5.5 GHz microwave photons. On-chip transduction efficiencies track well with predictions for single-photon EO coupling rates $g_0/(2\pi) \sim 1$ kHz. The transduction bandwidths are $15\ \mathrm{MHz}$ in both directions, sufficient for multiplexed quantum networking protocols. Notably, the static DC bias required for optical-microwave phase alignment is stable over multiple days of operation, obviating the need for active compensation even at cryogenic temperatures—a significant operational simplification over TFLN-based platforms.

Optical and Microwave Losses

Fabricated optical resonators demonstrate propagation losses as low as $0.25\ \mathrm{dB/cm}$. Wafer-scale statistics reveal well-controlled internal losses (e.g., post-process transducer resonators centered at 229 MHz linewidths), incurring only minor degradation from metallization. The absence of PECVD cladding contributes to the elevated microwave coherence—though the observed internal quality factors $Q_{\text{int}} \sim 1,100$ remain below expectations set by material loss tangent bounds, indicating the presence of residual, possibly process-induced, loss mechanisms. The extracted microwave participation ratios suggest improvement potential via substrate and oxide engineering.

Added Noise and Stability

Under pulsed operation with high peak optical power and low duty cycle, optically induced excess noise on the microwave resonator remains $<1$ photon per bandwidth for $100\,\mu$s pulses at maximal efficiency ($\sim 0.04\%$ on-chip). Notably, noise and bias drift correlate primarily with scattered light from grating couplers, not intrinsic EO processes, and can be further mitigated by optimizing coupler designs. No evidence of power-induced permanent degradation or thermalization lag is observed except for minor effects above $-20\ \mathrm{dBm}$ CW power.

Comparison with TFLN and Other Material Platforms

Compared to TFLN and other EO materials such as AlN and BaTiO${}_3$, TFLT provides several key advantages:

• DC Bias Stability: The platform supports operation with static biases, in contrast to the dynamic compensation required for TFLN due to spontaneous polarization drift.
• Process Scalability: DUVL enables order-of-magnitude higher device yields per wafer than e-beam lithography-based TFLN approaches.
• Material Stability: The larger bandgap and suppressed photorefractive effects permit higher optical pump intensities and operational stability over extended timescales.
• Loss Tangent: The air-clad TFLT geometry suppresses microwave dielectric losses below achievable limits in oxide-clad or substrate-limited geometries.

The demonstrated $g_0/(2\pi)\sim1$ kHz is comparable with state-of-the-art TFLN [25], but the operational and fabrication advantages point to more readily scalable quantum interconnect implementations.

Implications for Quantum Networks and Modular Architectures

TFLT-based EO transducers show particular promise for modular quantum processors interconnected via low-loss fiber optical links. The observed robustness under continuous operation, together with sub-photon added noise transduction for realistic pulse protocols, suggests suitability for distributed entanglement generation protocols, such as time-bin encoded Bell state generation. Theoretical modeling with projected platform enhancements (optimized optical $15\ \mathrm{MHz}$0, lower coupler loss, increased $15\ \mathrm{MHz}$1) indicates the potential for heralded entanglement fidelity up to 99.9% at rates exceeding $15\ \mathrm{MHz}$2 kHz, with minimal dissipation. When off-chip coupling losses are minimized, the TFLT architecture supports operating regimes compatible with high-fidelity modular quantum communication and distributed quantum memory protocols.

Pathways for Performance Enhancement

The demonstrated platform offers several avenues for further improvement:

• Resonator Loss: Process optimization (sidewall smoothing, debris mitigation) can push optical losses toward the $15\ \mathrm{MHz}$3 MHz regime, significantly increasing the power efficiency of transduction.
• Microwave $15\ \mathrm{MHz}$4: Deeper investigation into processed LT microwave loss mechanisms, substrate oxide participation, and improved air-bridge placement can unlock higher $15\ \mathrm{MHz}$5, increasing conversion bandwidth and fidelity.
• Coupler Engineering: Transitioning from grating to edge couplers, photonic wirebonds, or evanescent couplers is expected to reduce coupling loss below $15\ \mathrm{MHz}$6 dB, further diminishing noise contributions and increasing total system efficiency.
• Compact Design: Reduced cavity volumes will increase $15\ \mathrm{MHz}$7, improving transduction at lower pump powers and facilitating dense integration.

Conclusion

The realization of stable, bidirectional, and efficient EO transduction in TFLT integrated devices validates the material as a scalable and operationally robust platform for quantum interconnections. The demonstrated bias stability, wafer-scale process yield, and low added noise position TFLT as a strong alternative to TFLN for large-scale, modular quantum network deployment. Further improvements in optical loss, microwave coherence, and coupling strategies are expected to enhance the transduction efficiency, bandwidth, and scalability, supporting the practical implementation of photonic quantum state transfer and entanglement over optical links. TFLT platforms are thus poised to serve as foundational components in the architecture of future distributed quantum processors and photonic network infrastructures.