
Engineering and Technology
Lithium tantalate photonic integrated circuits for volume manufacturing
C. Wang, Z. Li, et al.
This innovative research by Chengli Wang and colleagues showcases the potential of lithium tantalate (LiTaO3) in developing low-loss electro-optical photonic integrated circuits (PICs), presenting a breakthrough alternative to traditional lithium niobate (LiNbO3) materials. Their findings reveal impressive properties for a 5G filter application, promoting scalable and cost-effective solutions for the future of communication technologies.
Playback language: English
Introduction
Next-generation ultrahigh-speed photonic integrated circuits (PICs) based on electro-optical materials are crucial for energy-efficient data centers, optical communications, and AI-driven high-performance computing. However, widespread adoption requires scalable and low-cost manufacturing. Silicon photonics, driven by the vast microelectronics industry, demonstrates the importance of high-volume, cost-effective wafer availability. Lithium niobate-on-insulator (LNOI) technology offers high-speed, low-voltage electro-optical PICs, but its commercialization is hampered by the lack of a large consumer electronics market driving its demand. In contrast, lithium tantalate (LiTaO3), another ferroelectric material with similar properties to LiNbO3, is already used extensively in 5G filters, allowing for large-scale, low-cost manufacturing. This study investigates the potential of LiTaO3 for creating PICs, leveraging its established manufacturing infrastructure and advantageous material properties. Prior research in LNOI has revealed significant advances in ultrahigh-speed, low-voltage electro-optical PICs, which could revolutionize communication systems. Yet, the high cost and limited wafer size of LNOI present obstacles to commercialization. Unlike silicon-on-insulator (SOI) technology which benefits from the massive scale of the microelectronics industry, LNOI currently lacks such a driving force. Lithium tantalate-on-insulator (LTOI) offers a potential solution, with an expected production capacity of 750,000 wafers annually by 2024, driven by its successful integration into 5G radio frequency filters. LTOI exhibits comparable or in certain instances superior properties to LiNbO3, including increased strength, chemical stability, and a wider optical bandgap, making it a prime candidate for a high-volume, cost-effective electro-optical PIC platform.
Literature Review
Existing research extensively documents the potential of LiNbO3-based PICs for high-speed modulation and various applications. Studies highlight the impressive capabilities of LiNbO3 in creating linear and high-speed modulators operating at CMOS-compatible voltage levels, suitable for data center communications, high-performance computing, and AI accelerators. However, the high cost associated with LiNbO3 wafer production and the limited wafer size significantly restrict its large-scale industrial application. The success of silicon photonics serves as a compelling example of how high-volume manufacturing significantly impacts the cost and feasibility of a technology. The smart-cut technique, commonly used in silicon wafer fabrication, has also been adapted for LiNbO3, producing LNOI structures. However, the lack of a large consumer market for LNOI hinders cost reduction. Previous work demonstrated the benefits of LTOI for radio frequency filters in 5G technology, and this study uses this commercial platform for high-volume fabrication and explores its potential in the domain of photonics. The optical properties of LiTaO3 are documented to be comparable, or even superior in some cases, to LiNbO3, further supporting its suitability for PIC fabrication. While free-standing LiTaO3 resonators have been fabricated using techniques like femtosecond laser direct writing and focused ion beam milling, a scalable manufacturing process for PICs remained elusive until this study.
Methodology
The fabrication of LTOI wafers employed the smart-cut technique, but unlike LNOI fabrication using high-energy helium ions, it utilized hydrogen ions at lower energy and higher beam current, aligning with commercial ion implanter capabilities. This resulted in a 4-inch LTOI wafer with a smooth surface (0.25 nm roughness) and consistent thickness. High-resolution scanning transmission electron microscopy confirmed the high quality of the LiTaO3-SiO2 interface. Photonic components, including ring resonators, racetrack resonators, and waveguide spirals, were fabricated using optimized lithography, dry etching, and cleaning processes to achieve effective coupling regions and well-defined sidewalls. The removal of non-volatile by-products from LTOI etching required a different chemical process compared to LNOI. Optical characterization was performed using frequency-comb-calibrated tunable diode laser spectroscopy to determine optical loss and absorption. The contributions of optical absorption and scattering were separated using thermal response spectroscopy. Electro-optical modulation was demonstrated using a tunable high-Q microresonator with DUV-lithography-based electrodes. The voltage tuning coefficient and resonance position were measured. A traveling-wave Mach-Zehnder modulator (MZM) was fabricated and characterized for electro-optical bandwidth and VπL. Finally, soliton microcomb generation was investigated, taking into account the polarization-dependent Raman effect in LiTaO3. The orientation of the racetrack microresonators was optimized to minimize Raman interference and facilitate soliton generation. A rapid single-sideband tuning scheme was used to initiate solitons, and the generated microcombs were characterized using optical spectrum analysis and fast photodiode detection.
Key Findings
The study achieved low propagation loss (5.6 dB m⁻¹) in LiTaO3 PICs fabricated using a DUV-based process. A LiTaO3 Mach–Zehnder modulator demonstrated a half-wave voltage–length product of 1.9 V cm and an electro-optic bandwidth exceeding 40 GHz. The low birefringence of LiTaO3 (more than an order of magnitude lower than LiNbO3) enabled the fabrication of thick waveguides without mode mixing, ensuring broadband operation across all telecommunication bands (1260 nm to 1625 nm). The platform supported the generation of soliton microcombs, with single solitons achieved at repetition rates of 81 GHz and 30.1 GHz. The 81 GHz soliton exhibited a 4.9 THz bandwidth, corresponding to a 63 fs pulse duration. The phase noise of the 30.1 GHz microwave beat note was measured, demonstrating its potential for high-quality microwave generation. Thermal response spectroscopy revealed that scattering losses were the dominant source of loss in the tightly confining LiTaO3 waveguides. Importantly, the absorption-limited propagation loss was found to be 0.4 dB m⁻¹, which is on par with results recently achieved in LNOI. The significant reduction in Raman intensity when the pump light polarization was aligned along the non-polar axis demonstrated that the polarization orientation was a key aspect in generating solitons in LiTaO3. In the optimized configuration, solitons were consistently observed across ten devices, while no soliton generation was achieved in devices oriented to maximize Raman interference. This confirmed the effectiveness of the proposed approach to mitigate the Raman effect.
Discussion
The results demonstrate the successful fabrication of low-loss, high-performance LiTaO3 PICs using a scalable and cost-effective manufacturing process. The achieved performance metrics, including low propagation loss, high electro-optical modulation efficiency, and soliton microcomb generation, are comparable or superior to state-of-the-art LNOI devices. The low birefringence of LiTaO3 is a crucial advantage, enabling the fabrication of compact, high-density circuits with broadband operation across all telecommunication bands. The demonstrated soliton microcomb generation in the x-cut configuration, previously unattainable in LiNbO3, further highlights the unique advantages of LiTaO3. The successful mitigation of the Raman effect by controlling the crystal orientation opens new possibilities for soliton-based applications. The use of a commercially established material and manufacturing process for 5G filters significantly reduces the cost and complexity of PIC production, making it a practical and attractive alternative to LNOI.
Conclusion
This work establishes a scalable manufacturing process for high-performance LiTaO3 PICs. The low-loss waveguides, high electro-optic modulation efficiency, and successful soliton microcomb generation demonstrate the potential of LiTaO3 as a platform for next-generation photonic integrated circuits. The leverage of existing high-volume manufacturing infrastructure for 5G filters significantly reduces the cost barrier for widespread adoption in various applications, including data center interconnects, long-haul optical communications, and quantum photonics. Future research could focus on optimizing waveguide designs, exploring new functionalities, and further improving the performance of LTOI-based devices.
Limitations
While this study demonstrates significant progress in LiTaO3 PIC fabrication, certain limitations remain. The phase noise of the generated microwave beat note is higher than some reported values for other platforms. Further optimization of the fabrication process and device design could potentially reduce scattering losses and improve the phase noise characteristics. Additional research is needed to fully explore the potential of LiTaO3 for high-power applications given its superior optical damage threshold. The current study focuses primarily on specific functionalities; a comprehensive comparison across a broader range of applications would further solidify the position of LiTaO3 in photonic integrated circuits.
Related Publications
Explore these studies to deepen your understanding of the subject.