Photonic integrated circuits have driven advances in optical interconnects, wireless communications, seamless sensing. Electro-optic modulators, which bridge the electronic and optical domains, act as the cornerstone of these systems. To address the ultra-broadband demands of next-generation information systems while enabling heterogeneous integration with other functional components, here we present a high-performance thin-film lithium tantalate-on-silicon nitride heterogeneous Mach-Zehnder modulator fabricated via micro-transfer printing. The device shows an insertion loss of 1.0 dB, an electro-optic bandwidth exceeding 110 GHz, and a power drift of ~0.3 dB over 72 h at quadrature bias. We employ the modulator for three typical scenarios in the applications of optical communications, fully photonic terahertz wireless communications, and radar sensing. It achieves optical data transmission rates beyond 375 Gbps, a record rate of 148 Gbps for wireless communications in the D-Band, and a range resolution of 1.2 cm with measured errors below 2 mm in the W-Band. These results pave the way for scalable, high-performance, cross-domain photonic chips for next-generation information networks. Researchers demonstrate a micro-transfer-printed lithium tantalate-on-silicon nitride modulator with 110 GHz bandwidth and 1 dB loss, enabling high-speed optical and terahertz communications and high-resolution radar sensing.

Beamforming technology featuring broad bandwidth, fast beam switching speed and multi-user access capability plays a pivotal role in the next-generation wireless communication systems. Here, we report a fully-connected broadband microwave photonic multi-beamforming architecture based on silicon optical true-time delay lines (OTTDLs). Unlike conventional partially-connected schemes, the fully-connected network allows each radio frequency beam to be synthesized by all antenna elements, enhancing beamforming gain and multi-beam scalability. Experimental measurements demonstrate squint-free beam steering across 13 directions with deflection angles exceeding ±50.0° and a fast beam reconfiguration speed of 60 ns. Compared with partially-connected schemes, our architecture achieves stronger received signals, delivering a 2.9 dB gain at 13.1 GHz for an identical 8-element antenna array. A 64-QAM downlink signal with 400 MHz bandwidth achieves an error vector magnitude (EVM) of 7.6% over 20-m outdoor wireless transmission. These results provide a scalable solution for broadband microwave photonic multi-beamforming and support future 5 G and 6 G wireless networks.

We demonstrated an 850 nm external cavity laser with a wide tuning range of 33 nm and an output power of 11 dBm, enabled by a gain chip integrated with a low-loss Si 3 N 4 feedback chip.

We present a lens-free, solid-state structured light imaging system utilizing a 1×64 optical switch array of large-aperture slanted grating antennas. The proposed design removes the constraints of conventional lens-assisted or optical phased arrays illuminators, enabling a large field of view of 100°×8° without limitations concerning active optical alignment or aliasing side lobes.

A fast and low-cost pre-distortion linearization algorithm for laser frequency sweeping calibration is highly desired for frequency-modulated continuous wave (FMCW) lidar applications. We propose a frequency resampling-based pre-distortion strategy, which enables the laser sweeping residual nonlinearity to reach 10-8 level with only 8 or less iterations and does not require any additional hardware compared to external modulation or optical phase locked loop methods.

We report high-performance frequency-modulated continuous wave (FMCW) sources for high-resolution and high-precision detections based on narrow-linewidth III-V/Si3N4 hybrid integrated external cavity lasers (ECLs). Through synchronously tuning the cavity filter and the phase shifter, a highly linear chirp with 21.8 GHz bandwidth is generated at 1 kHz repetition rate, allowing for a ranging resolution of 1.64 cm. Ranging over 10 wavelengths is demonstrated with the same modulation driving conditions, featuring an overall precision of 1.99 mm for a 10-m fiber. Furthermore, by employing a customized electro-optic phase-locked loop (EO-PLL) implemented on a printed circuit board for noise suppression, we achieve a highly linear frequency chirp with the nonlinearity reduced to 6.01x10(-7). A precision improvement from 4.44 m to 10.28 cm over a 300-m range is achieved with the EO-PLL lock-on.

A joint phase-amplitude controlled microwave photonic in-phase/quadrature (I/Q) mixer for image-rejected multichannel reception is proposed. A dual-parallel Mach-Zehnder modulator (DPMZM) performs carrier-suppressed double-sideband modulation on both upper and lower sub-MZMs. Phase control is implemented by adjusting the bias voltage of the phase shifter inside the DPMZM, enabling flexible and continuous tuning of the phase difference between the positive and negative sidebands of two local oscillator (LO) signals for I/Q mixing. Amplitude control is realized by setting the phase difference between the two LO signals, allowing reconfigurable amplitude responses of the I- and Q-branches. A set of LO signals is configured with discrete phase differences to establish a mapping relationship between the amplitude response ratios and the sub-bands, thereby enabling a multichannel I/Q mixer. Benefiting from this joint phase-amplitude control mechanism, I/Q mixing across multiple sub-bands is achieved without the need for LO switching. A proof-of-concept experiment is demonstrated. The single sub-band I/Q mixer operates over a 5-40 GHz frequency range, achieving a phase mismatch below 2 degrees, an image rejection ratio (IRR) exceeding 35 dB, and a spurious-free dynamic range of approximately 101.27 dB center dot Hz(2/3). A multichannel I/Q mixer is implemented at three sub-bands, achieving a phase mismatch below 4 degrees and an IRR exceeding 28 dB, and a six-frequency hopping experiment without switching time is further demonstrated, achieving accurate recovery of the hopping frequencies.

Objective With the increasing demand for integrated optical frequency comb (OFC) sources, research into integrated mode-locked lasers (MLLs) has intensified. Early integration efforts primarily focused on passively mode-locked lasers employing saturable absorbers. However, the stable operating range of such devices is typically confined to narrow windows of driving current and bias voltage. In contrast, actively mode-locked operation achieved through the hybrid integration of reflective semiconductor optical amplifiers (RSOAs) and thin-film lithium niobate (TFLN) external cavities offers significantly higher tunability in terms of both output power and repetition rate. An early study of RSOA-TFLN integrated platforms demonstrates mode-locked operation, but the achievable spectral bandwidth remains in the sub-nanometer regime. Although subsequent research attempt to broaden the spectrum by introducing the Kerr effect within a microresonator, such methods necessitate dispersion engineering and partially compromise the tuning flexibility associated with harmonic mode locking. To address the spectral bandwidth limitations of previous actively mode-locked lasers, this work exploits the broadband gain of the RSOA to broaden the mode-locked spectrum and demonstrates a broadband microcomb mode-locked laser based on a simple integrated configuration, providing an effective pathway toward chip-scale optical frequency combs. Methods Active mode locking relies on periodic phase modulation to establish phase synchronization among longitudinal modes within the resonator. When the modulation frequency matches the cavity free spectral range (FSR), equally spaced sidebands enable effective coupling between adjacent modes, leading to stable mode-locked operation. The laser is described by the Haus master equation (HME), which accounts for the combined influences of gain, loss, cavity dispersion, and modulation depth. By assuming a steady-state solution in the form of a chirped Gaussian pulse, an analytical expression for the spectral bandwidth is derived, revealing the combined influence of gain bandwidth, dispersion, and modulation depth on the spectral performance of the mode-locked laser. Experimentally, an actively mode-locked laser is realized through hybrid integration of an RSOA chip and a TFLN chip. The TFLN chip is fabricated on a 600-nm-thick X-cut MgO-doped lithium niobate-on-insulator wafer, with ridge waveguides of 300 nm depth defined by inductively coupled plasma dry etching. Gold traveling-wave electrodes with a thickness of 800 nm are fabricated by electron-beam evaporation, forming a 6.5-mm-long phase modulator. To minimize intracavity loss, a spot-size converter (SSC) with a tapered structure is integrated on the TFLN chip to expand the waveguide width from 1.4 mu m in the modulation region to 5 mu m at the coupling facet, ensuring optical mode matching with the RSOA. Finite-difference time-domain simulations indicate a theoretical coupling loss of 1.33 dB. Further analysis of the coupling loss as a function of waveguide width and alignment offsets reveals that, owing to the small mode-field height, vertical alignment precision is critical for maintaining high coupling efficiency. The mode-locked laser cavity is formed by the high-reflection-coated facet of the RSOA and an integrated Sagnac loop mirror on the TFLN chip. Results and Discussions To characterize the gain properties of the RSOA, its amplified spontaneous emission (ASE) spectrum is measured in the absence of external cavity feedback. At a driving current of 100 mA, the RSOA exhibits a broad ASE spectrum with a 3 dB bandwidth of approximately 70 nm, covering the wavelength range from 1450 to 1650 nm. Periodic oscillations observed at the top of the ASE spectrum originate from residual reflections at the facet of the tapered optical fiber. After coupling the RSOA to the TFLN chip, the laser output without a raio frequency (RF) driving signal reveals a random multi-longitudinal mode distribution characterized by large intensity fluctuations and the absence of a stable spectral envelope. When an RF driving signal with a frequency matched to the cavity FSR is applied (6.17 GHz, 33 dBm), the laser transitions into a stable actively mode-locked state. The resulting output spectrum exhibits a remarkably flat and broad envelope with a 3-dB bandwidth of 5.5 nm. The optical spectrum above the noise floor spans from 1589 to 1604 nm, corresponding to a total of 283 comb lines, among which 105 lines fall within the 3 dB power window. A magnified spectral view confirms that the comb-line spacing is strictly equal to the applied RF modulation frequency of 6.17 GHz. The coherence of the generated frequency comb is further evaluated by measuring the beat-note signal using an electrical spectrum analyzer. With a resolution bandwidth of 100 Hz, the beat note exhibits a signal-to-noise ratio of 62 dB, confirming strong phase coherence among the comb lines and stable phase synchronization of the longitudinal modes. Conclusions In conclusion, this research provides a systematic theoretical analysis and experimental demonstration of an actively mode-locked laser based on a hybrid III-V/TFLN platform. Analysis based on the HME indicates that broadband gain is the fundamental prerequisite for broadband spectral expansion. By integrating a broadband RSOA chip with a high-speed TFLN external cavity, a mode-locked microcomb laser with a 6.17 GHz repetition rate and a 5.5 nm 3 dB spectral bandwidth is demonstrated. These results underscore the immense potential of the integrated III-V/TFLN platform for providing high-performance on-chip frequency combs for optical communications, LiDAR, and precision metrology.

The rising computational demands of deep learning-particularly for high-order tensor operations-have driven the need for novel hardware accelerators. While integrated photonics presents a promising platform, the matrix scale of existing schemes is often constrained by the physical number of on-chip components. Here, we propose and experimentally demonstrate a scalable high-order integrated photonic tensor processor (HIPTP) that decouples the computational dimensionality from physical device count by leveraging a two-dimensional array combined with electrical frequencydomain modulation. The design employs a scalable array of cascaded Mach-Zehnder interferometers (MZIs) and photodetectors (PDs), enabling third- and fourth-order tensor multiplication by encoding data onto distinct microwave frequencies. A 4 & times; 4 HIPTP chip was fabricated and characterized, exhibiting high consistency across core components, which facilitates the extension toward larger-scale on-chip systems. To validate its computational capabilities, three computational modes were verified under varying frequency counts, and an inference accuracy of 97.5% was achieved on the German traffic sign recognition benchmark (GTSRB). Our work presents a promising pathway toward ultra-largescale photonic computing for advanced AI models and frequency-aware signal processing scenarios. (c) 2026 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

This study proposes and demonstrates a silicon-integrated 6-bit tunable optical true-time delay line (OTTDL) based on high-extinction-ratio (ER) electro-optic (EO) switches with fast tuning speed. Each switch is composed of a push–pull PIN-based Mach–Zehnder interferometer (MZI) switch structure with a tunable coupler to balance interference, which enhances the ER to >34 dB. The delay line features a delay step resolution of 5 ps and a maximum delay of 315 ps. The delay ripple of the OTTDL is suppressed to below 0.8 ps over a broad bandwidth of 10–43.5 GHz. With adequate switch ER, signal leakage suppression is achieved without activating variable optical attenuators (VOA), lowering the total power consumption to 166 mW.

Chip-scale optical frequency combs (OFCs) enable integrated solutions for various applications among optical communications, LiDAR, spectroscopy, and radio frequency signal generation. Among integrated OFC platforms, III-V semiconductor mode-locked lasers provide gigahertz repetition rates but often suffer from limited power and tunability due to narrow driving current ranges. Hybrid integration with thin-film lithium niobate (TFLN) has been explored to address these limitations, but previous III-V/TFLN mode-locked lasers exhibited narrow spectral widths, constrained by gain bandwidth and cavity dispersion. This work demonstrates broadband III-V/TFLN integrated actively mode-locked lasers at the telecom band, leveraging a broadband multiple-quantum-well reflective semiconductor optical amplifier (RSOA). A mode-locked laser with a high-reflectivity TFLN mirror generates a 45-nm-span spectrum with a 10-dB bandwidth of 19.1 nm and a beat-note contrast exceeding 60 dB, while another laser with a low-reflectivity RSOA output facet achieves an off-chip output power of 26 mW with a 10-dB bandwidth of 11.4 nm. Numerical simulations based on the Haus's master equation and the generalized nonlinear Schr & ouml;dinger equation confirm the role of slight inhomogeneous gain in supporting stretched-state pulse solutions and overcoming dispersion-limited spectral narrowing. This work demonstrates outstanding spectral performance and provides a promising route toward broadband, high-power OFC sources for integrated photonic applications.

Optical frequency transfer via fiber-optic networks achieves remarkable precision, serving as a critical enabler for the development of optical clock networks. Neverthe less, in long-distance fiber links, polarization mode dispersion (PMD)—resulting from birefringence—is highly vulnerable to environmental disturbances such as temperature variations and mechanical vibrations, leading to increased long-term system instability. In this work, we experimentally propose and demon strate a PMD suppression scheme based on synchronized polarization scrambling and descrambling in a two-way optical frequency comparison (TWC) system. A theoretical model is established to quantify the temperature-dependent PMD-induced delay error, and we introduce a 10-MHz clock distribution over the same fiber link to synchronize the scramblers at both ends, ensuring long-term effective descrambling. Experimental results show that, with scrambling, the temperature sensitivity of the TWC phase is reduced from 9.85 as/K to 1.34 as/K. The overlapping Allan deviation is improved from 1.32×10−19 to 6.22×10−20 at 1,000 s. This approach effectively suppresses the temperature-induced long-term instability without introducing additional phase noise, providing a practical solution for robust large-scale optical clock networks.

Optical computing leverages high bandwidth, low latency, and power efficiency, which is considered as one of the most effective solutions for accelerating deep learning tasks. However, mainstream photonic hardware accelerators are primarily optimized for two-dimensional (2D) matrix-vector multiplications (MVMs). To implement three-dimensional (3D) convolutional neural networks (CNNs), high-order tensors must be reshaped in the electrical domain according to the size of the accelerators before computation, leading to extra memory usage and time overheads. Additionally, synchronization across multiple channels depends on external electronic clocks, which increases the complexity of the system. In this work, we propose an integrated photonic 3D tensor processing engine (3D-TPE) based on the interleaving modulation of time, wavelength, and space. Data caching, channel synchronization and computation are realized entirely within the optical domain, reducing memory and time usage, and simplifying the system. Optical caching and synchronization are achieved with an optical tunable delay line (OTDL) chip supporting versatile clock frequencies up to 200 GHz, and optical computing is accomplished with a dual-coupled micro-ring resonators (MRRs) based crossbar chip with a 3-dB passband width of 50 GHz. We verify the processing capabilities of the 3D-TPE at clock frequencies ranging from 10 GHz to 30 GHz and perform a proof-of-concept experiment for a LiDAR 3D point cloud image recognition task operating at 20 GHz, achieving a recognition accuracy of 97.06%. The proposed 3D-TPE is anticipated to facilitate high-order tensor convolutions, playing an important role in autonomous driving, healthcare, video analytics, virtual reality, etc.

Network architectural is indispensable for a wide-area and robust Fiber-optic time transfer (FOTT) network, however, there is currently a deficiency in effective scheme. This article proposed a reconfigurable FOTT network architecture based on software-defined time transfer units (SD-TTU), enhancing the network's capability for flexible resource allocation, automatic protection and restoration. The architecture consists of three layers: 1) time transfer layer for time signal transmission and clock difference cancellation; 2) monitor and control layer for monitoring network states, fault detection, and dynamic reconfiguration of the SD-TTU; and 3) service and configuration layer for customizable service construction and proactive network reconfiguration. A polymorphic software-defined time transfer unit (SD-TTU) is further designed, which can operate as a master station, intermediate station, and slave station using the same hardware. The proposed architecture and the designed SD-TTU are experimentally demonstrated over chained and ring networks, respectively. Network expansion, contraction, topology reconfiguration, failure and fault recovery are validated. The changes of measured mean clock differences are below 43-ps across different scenarios, verifying the robustness and flexibility of the proposed architecture.

We demonstrate a high-performance III-V/Si3N4 mode-locked laser. Stable passive mode-locking achieves a wide 3-dB bandwidth of 9.12 nm and a narrow 10-dB linewidth of 443 Hz. Hybrid mode-locking further enhances the frequency stability.

We demonstrated a fast polarization-insensitive optical switch on the 220-nm SOI platform with an insertion loss of 0.92/2.79 dB and a response time of 52.2/55.0 ns for TE0/TM0 mode at the 1550 nm wavelength.

We propose a physically realizable temporal tail-cutting algorithm for Ge-Si photodetectors. We achieve a 160 Gbps data transmission with the transmitter dispersion eye closure quaternary (TDECQ) enhanced from 5.13 dB to 2.64 dB, when the algorithm is engaged.

Long-distance fiber-optic time and frequency transfer (FoTFT) techniques are fundamental for a vast number of applications. To compensate the signal attenuation caused by the fiber loss, current FoTFT techniques generally use optical amplifiers or/and repeaters to boost or regenerate the transferred signals. However, the use of optical amplifier or/and repeater face several problems, including the deterioration of signal-to-noise ratio, the increase of system complexity, etc. Here, we demonstrate a high-sensitivity time and frequency demodulation algorithm, which can enable a long-span fiber-optic time and frequency transfer without using optical amplifiers or/and repeaters. A receiving sensitivity of -90 dBm is achieved in experiment. The precision of time and frequency demodulation is 5.54 ps and 7 mHz at -40 dBm, and 1755 ps and 2.9 Hz at -90 dBm, respectively.

Platicon microcombs have been demonstrated on various material platforms, offering wide spectral bandwidth, excellent spectral flatness, and efficient power conversion, which are of interest for numerous applications. Lithium niobate (LN) with its Pockels and Kerr effects has become a promising platform for both generating and controlling microcombs. The photorefractive (PR) effect intrinsic to LN significantly influences the dynamics of Kerr frequency comb generation in microresonators. Here, we experimentally and theoretically analyze platicons in a LN microresonator, demonstrating that the PR effect promotes comb generation during scans from the red-detuned side but hinders it when scanning from the blue-detuned side. We also report platicon microcombs generated in normal-dispersion LN microresonators, which span over 250 nm and achieve a conversion efficiency of 13.5%. This Letter expands the understanding of LN microresonators with the PR effect and highlights their potential as platforms to develop highly efficient and flat optical comb sources.