Jicong Zhao

The long-wave infrared (LWIR) band lies near the peak of natural thermal emission, enabling stable passive imaging without active illumination, and it also covers characteristic spectral features of many important materials, making it a key spectral region for sensing and identification. However, the development of micro-electromechanical systems Fabry–Pérot tunable filters (MEMS-FPTFs) in this band is limited by strong material absorption, scarce low-loss thin-film options, and the difficulty of fabricating thick, low-stress, high-reflectivity distributed Bragg reflectors (DBRs). This work presents an electrostatically actuated MEMS-FPTF tailored for LWIR operation through an optical–mechanical co-design strategy. A low-loss Ge/ZnS DBR with > 92% reflectivity, combined with a C₃-symmetric three-terminal ring suspension and a bulk-micromachining fabrication flow (C₃ denotes threefold rotational symmetry with a 120° periodicity), enables high spectral finesse and stable large-stroke actuation. The device achieves a minimum full width at half maximum (FWHM) of 188 nm, a maximum quality factor (Q) of 63.7, and continuous tuning from 10.3 to 12 μm, representing advanced performance among LWIR MEMS-FPTFs. To further demonstrate application potential, we developed a dual-channel spectral–visual proof-of-concept recognition demonstration based on the MEMS-FPTF. Compared with a vision-only neural-network baseline, the hybrid model exhibits a significant improvement in recognition performance under illumination variations and complex background conditions. The proposed MEMS-FPTF offers a compact, low-power, and high-precision solution for LWIR spectral filtering, with strong promise for mineral identification, autonomous perception, and remote spectral sensing.

In this letter, we demonstrate an effective route to suppress transverse spurious modes in A1-mode LiNbO3 Lamb wave resonators by shaping the busbars into a crescent profile. The resulting crescent-shaped excited bulk acoustic wave resonator (CS-XBAR) introduces a quadratic phase variation along the transverse direction, which establishes an energy gradient and drives destructive interference of transverse leakage waves at the busbar boundaries. A waveguide model shows that the suppression of higher-order transverse modes is maximized when a triple-busbar configuration is used and the curvature radius matches the resonator period. Devices are fabricated on a Z-cut LiNbO3-on-Si platform and exhibit a resonance frequency of 4.45 GHz with an electromechanical coupling coefficient kt2 of about 27.2%, a maximum Bode-Qmax of 1321, and a spurious-free admittance spectrum over the measured band, yielding a figure of merit kt2 × Bode-Qmax of about 360. Compared with co-fabricated straight single-busbar references, the CS-XBAR shows markedly enhanced spectral purity and is promising for low-loss N79-band 5G RF filters.

This study reports AlSc0.095N Lamb-wave resonators (LWRs) using an interdigitated (IDT)–IDT transducer and integrated transverse reflectors that simultaneously suppress transverse spurious modes and mitigate anchor loss. The influence of the electrode metal ratio (η) on the resonators’ frequency, quality factor (Q) values, and effective electromechanical coupling coefficient (kt2) was explored through theoretical analysis, finite-element simulations, and measurements. To enable crack-free AlScN above patterned electrodes, we develop an 8-in. wafer-level transfer flow (Si–SiO2 bonding) that decouples AlScN deposition from bottom-IDT patterning and improves film crystallinity. Placing Mo reflectors around the bus and tether regions forms a low-acoustic-velocity frame that impedes lateral standing-wave build-up and reflects leakage at anchors, removing the parasitic responses between series and parallel resonance peaks. Prototypes operating near 412.8 MHz exhibit spurious-free spectra, Bode-Qmax up to 2387, and kt2 of 2.8%, yielding FoM = Bode-Qmax⋅kt2 = 66.83.

We first propose and demonstrate a compact and high sensitivity vector curvature sensor based on a dual-path Mach-Zehnder interferometer (MZI) structure with intensity-modulated operation. The sensor is fabricated via electric-arc splicing technology using two parallel optical paths, where the reference path is composed of a single-mode fiber (SMF) while the sensing path incorporates the SMF with a core-offset fiber structure. The asymmetric arrangement of the dual-path MZI, combined with the core-offset fiber design, creates a direction-dependent curvature sensitivity characteristic. Experimental results demonstrate that the proposed sensor can distinguish four directions: 0°, 90°, 180°, and 270°. For the bending direction of 180°, a peak sensitivity of −105.54 dB/m^-1 is achieved over the range of 0 to 0.041796 m⁻¹, which is the highest sensitivity for intensity-modulated vector curvature sensors to our knowledge. Additionally, we conduct experiments on temperature and strain. The experiments show that the proposed sensor exhibits temperature and strain sensitivities of 86 pm/°C and 4.3 pm/με, respectively, within the ranges of 30 °C–90 °C and 240–860 με. The cross-sensitivity issues of these parameters can be effectively addressed by a sensitivity matrix. The dual-path MZI structure, designed for vector detection, illuminates vision for developing sensors with various vector characteristics.

This paper presents a laterally excited bulk acoustic resonator (XBAR) with reflective arrays, enabling the effective suppression of lateral acoustic leakage and enhanced energy confinement. Building upon previous studies, this work further investigates the critical design parameters of reflective arrays, specifically the optimal distance between the reflective arrays and the interdigital transducers (IDTs) and the selection of the reflective-array metal coverage at this optimal distance. Meanwhile, by systematically analyzing the key parameters of IDTs, including electrode pitch and electrode numbers, their influences on the quality factor (Q value) and spurious modes were also studied. Comparative studies with conventional XBARs and full-release air boundary XBARs further highlight the superior structural stability and performance advantages of the reflective-array design. Both simulations and experimental results confirm that the optimized configuration leverages acoustic impedance mismatch to achieve efficient reflection, effectively improving the Q value. The fabricated devices demonstrate a 2.83× increase in Bode-Q from 279 to 789 while maintaining a high electromechanical coupling coefficient (kt2) of 39.1%. The optimized XBAR further achieves excellent figures of merit FOM1 (=kt2 × Bode-Q) of 308 and FOM2 (=fs × kt2 × Bode-Q) of 10.83 × 1011 Hz, underscoring its strong potential for 5G n77-band applications.

The rapid development of the Internet of Things is driving advances in neuromorphic photonic computing systems that leverage light for highly efficient and parallel information processing. Note that to achieve neuromorphic computing with artificial optical synapses, tunable synaptic plasticity and photo-memory functions enabled by direct optical modulation are essential. In this study, we design an organic optical synaptic device with a two-dimensional (2D) metal oxide nanosheet as the charge storage layer, demonstrating bimodal synaptic plasticity and multilevel photo-memory. The precise layer-by-layer liquid deposition technology enables direct integration of 2D Cs_2.7W₁₁O₃₅ ^1.3− nanosheet films with controllable thickness into the device for charge storage. Through interfacial band engineering, a type-II heterojunction is established between the organic semiconductor channel and the inorganic nanosheet, enabling gate-voltage-free photogenerated charge transfer and storage. Under optical modulation, the device exhibits tunable synaptic behaviors ranging from short- to long-term plasticity and a multilevel photo-memory capability with a retention time of 10³ s. Then, we confirm the potential of the synaptic device for highly efficient neuromorphic computing, demonstrating a recognition capability with a high accuracy of 97.8%. Therefore, our device sheds new light on the development of optical computing systems.