Photonic modulators are fundamental photonic devices that use external energy to alter the characteristics of optical signals, primarily enabling precise control over physical parameters such as amplitude, frequency, and phase. Integrated photonic chips based on these modulators, leveraging their high bandwidth, high-speed signal processing capabilities, and extremely low power consumption, have become the core hardware platforms for future optical communication, interconnects, and computing. Technological advancements in this field have consistently attracted significant academic interest and R&D investment in recent years. While traditional research has focused on modulation speed, current demands for miniaturization and energy efficiency are gradually shifting the emphasis toward modulation efficiency. This article systematically elucidates the mechanisms for optimizing the energy efficiency of photonic modulators, with particular emphasis on a comparative analysis of phase modulators based on silicon, Ⅲ-V compounds, ferroelectrics, synthetic polymers, and graphene. Through multi-dimensional performance evaluation and mechanism analysis, it provides a reference for forthcoming innovations in efficient photonic modulation devices.

A high-voltage and high-frequency test link has been designed for planar SiC and GaN photoconductive devices. Considering the differing characteristic impedances of wide-bandgap semiconductor devices, PSpice software was used to simulate the circuit design of the test link, while CST software was employed to simulate the high-frequency response within the 0.5~5 GHz range. The high-voltage conductivity and high-frequency response of the link were then tested using SiC and GaN photoconductive devices, respectively. Experimental results demonstrate that the proposed test link offers high voltage resistance, a fully solid-state design, and ease of disassembly. It meets the photoconductivity testing requirements of planar SiC and GaN photoconductive devices under bias voltage of 0~30 kV and within a DC~1 GHz operating frequency range. These findings provide valuable design and experimental references for evaluating the high-voltage tolerance and high-frequency response characteristics of wide-bandgap semiconductor photoconductive devices.

To solve the issues of insufficient memory and low processing speed caused by excessive data volume in low-frequency calibration under existing calibration methods, a low-frequency accelerometer calibration system based on homodyne laser interference is developed. The error-compensation method is proposed by analyzing nonlinear errors, particularly the non-orthogonal phase error. Using the dynamic successive phase unwrapping method and the self-adaptive dynamic decomposition algorithm for data acquisition and processing, the sampling rate and data volume are reduced, and the calibration accuracy is ensured. Experimental results show that the system can achieve high-accuracy sensitivity calibration of the accelerometer in the frequency range from 0.1 to 80 Hz and satisfy the test requirements of low-frequency calibration.

This study focuses on designing and optimizing a 4H-SiC superjunction trench field-effect transistor structure (DP-SJ-UMOS) with two segments of different P-pillar concentrations. The device′s UIS testing circuit was analyzed and its fundamental operating principles determined. The UIS characteristics of this device structure were studied in detail using Sentaurus TCAD simulation software, where three methods were proposed to improve avalanche tolerance. Multiple epitaxial growth and high-energy ion-implantation techniques were employed in the drift region to form the upper and lower segments of the superjunction structure with different concentrations, enhancing the avalanche current path during avalanche breakdown. This modification reduced the current of the parasitic transistor during breakdown, effectively suppressing the activation of the parasitic transistor and improving avalanche tolerance. The experimental simulations indicate that the proposed structure, compared with conventional superjunction devices (Con-SJ-UMOS), achieves a 1.5% increase in peak current, with breakdown voltage and avalanche tolerance enhancements of 24.5% and a 0.9%, respectively.

CMOS image sensors are characterized by low power consumption, low cost, high integration, high response speed, and high temperature-measurement limit, and are widely used for measuring radiation temperature. During a field experiment, the movement temperature of a CMOS image sensor fluctuates depending on the environmental temperature, which consequently alters the background grayscale and response coefficient, thus affecting the quantitative ability of the CMOS movement and increasing the temperature-measurement error. We propose a universal calibration method to address this issue. By constructing a temperature-calibration model for the background grayscale and response coefficient, we achieve stable CMOS movement values under different environmental temperatures and improve the temperature-measurement accuracy. We simulate environmental temperature changes from 273.2 to 313.2 K in an external field using high- and low-temperature chambers and use a standard high-temperature blackbody to obtain test data for a CMOS image sensor on a 1 073.2 K target. Experimental results show that this correction method can reduce the maximum temperature difference recorded by the CMOS image sensor from 10.9 to 2.2 K in a 273.2~313.2 K environment.

This paper introduces a readout circuit for a frame difference sensor based on the 180-nm CIS process. The data compression feature of the circuit reduces data transmission bandwidth and storage pressure. The advantages and disadvantages of the two data output formats, viz. frame difference and address-event representation, which are used in dynamic vision sensors, are analyzed to design a readout circuit that can adaptively switch between data output modes. This circuit dynamically selects the optimal data compression method based on event density, thereby effectively reducing redundant data. The auto-zeroing technology is employed to reduce the offset voltage in the event encoding circuit. Simulation results show that the auto-zeroing technique reduces the impact of offset voltage on quantization results (from 38.77 to 1.185 mV), thus significantly improving the accuracy and stability of the circuit. Additionally, the frame rate can be dynamically adjusted between 375 and 1627 fps to satisfy the requirements of different scenarios. Experimental results indicate that the circuit exhibits outstanding preformation in sparse event scenarios, with a maximum video compression ratio of up to 98%, which significantly enhances the efficiency of data transmission and processing.

Polarimetric detection, a critical technical approach for characterizing light–matter interactions, has been widely applied in biomedical imaging, high-precision optical metrology, and military target identification in recent years. To address and fulfill the development requirements for integration and miniaturization of optoelectronic systems, an innovative full-Stokes polarimetric detector, based on the silicon-on-insulator (SOI) technology, is proposed in this study. In this design, a 500 nm-thick SOI device layer is selectively patterned to form a metasurface structure and simultaneously engineered as a pin photodetector. This approach enables monolithic integration that effectively addresses the critical limitations of conventional polarization detection systems, i.e., their bulky form factors and poor integration capability. The results demonstrate excellent device performance at the operational wavelength of 780 nm, where the chiral metasurface pixels exhibit circular dichroism of 42%, and the linear metasurface pixels achieve an extinction ratio of 10 dB. Notably, high-accuracy Stokes parameter reconstruction is accomplished, with average measurement errors of 0.30%, 0.85%, and 11.17% for the parameters S1, S2, and S3, respectively. This study provides a novel design strategy for integrated polarimetric systems as well as highlights the considerable application potential of metasurface technology in the field of optoelectronic integration.

Mid-wave infrared (MWIR) HgCdTe detectors operating at room temperature face challenges in enhancing signal-to-noise ratio (SNR) and dynamic range during signal readout owing to their low impedance and high background currents, which limit their practical applicability under noncooled conditions. This paper presents a light-to-frequency converter circuit architecture designed to address these limitations. The proposed circuit stabilizes the detector’s operating state using a low-offset operational amplifier and employs a current-to-voltage-to-current-to-frequency conversion process to reduce the high photocurrent into a range suitable for the current–frequency input stage (0 to 4.3 A). Additionally, the circuit effectively suppresses dark and background currents originating from the detector. Experimental results demonstrate that the integrated system achieves a dynamic range of 93 dB with a power consumption of only 27.3 mW, enabling room-temperature, large-dynamic-range readout for MWIR HgCdTe detectors.

A mid-infrared cavity is implemented herein on a valley photonic crystal structure with a honeycomb lattice via manipulation of the valley degrees of freedom. The resulting topological waveguide designed considering this structure exhibits remarkable unidirectional transmission characteristics, achieving near-unity forward transmission efficiency within the operating frequency range while effectively suppressing backward scattering. The cavity constructed on the basis of this design achieves a quality factor of 5.59×104 within the target frequency range. When compared to conventional photonic crystal cavities, this mid-infrared cavity exhibits a significantly lower inverse participation ratio that reaches as low as 1.6 and consistently remains below 2 within the target frequency range. This results in significant mode field distribution uniformity. Moreover, the mode distribution of the cavity remains stable even when structural defects are introduced.

To meet the application requirements of large area array complementary metal-oxide semiconductor (CMOS) image sensors (CIS), a ramp signal generator with slope and sample mode adjustment is presented in this paper. The design, simulation, and layout are implemented using a 90 nm (1.2 V/2.8 V) 1P5M CIS process. This circuit design has a simple structure, small area, adjustable slope. The slope amplitude is greater than 0.5 V, reset time is less than 70 ns, differential nonlinearity is +0.018 LSB/−0.012 LSB, and integrated nonlinearity is +0.37 LSB/−0.013 LSB, which meets the design and engineering application requirements of large array CIS.

A 64 × 64 silicon-based linear-mode avalanche photodiode (APD) module was designed and fabricated for three-dimensional (3D) laser imaging. The silicon-based APD focal plane array, with a back-illuminated N+-p–-P+ structure and 150-m pixel pitch, operates in the linear avalanche mode. The APD pixels use an on-chip microlens, a reflecting mirror, and a composite antireflective coating to improve the photon detection sensitivity in the NIR range. The constituent readout integrated circuit integrates a high-speed fronted amplifier, a time discrimination circuit, high-accuracy TDCs, a timing control module, and other functions on a single chip. The 64 × 64 silicon-based linear mode APD arrays can simultaneously detect laser pulses with good uniformity. Test results show that the threshold optical power detected at 905 nm is 20 nW; in addition, the time resolution, time range, and maximum frame rate are approximately 0.339 ns, 16 bits, and 5 kHz, respectively. This component meets the detection requirements for high-precision, high-speed, long-distance, non-scanning laser 3D imaging.

With the widespread application of infrared imaging technology in fields such as night vision surveillance and industrial inspection, there is increasing demand for integrated infrared imaging systems. This paper presents the design of an infrared image-processing System-on-Chip tailored for integrated infrared imaging applications. The chip integrates a CPU, an Image Signal Processing co-processor, and dedicated peripheral control interfaces, enabling the monolithic implementation of real-time functions, including infrared detector control, image acquisition, adaptive non-uniformity correction, adaptive blind pixel recognition and compensation, and image enhancement. This integration significantly simplifies the architecture of the infrared detection signal processing system, reduces power consumption and cost, and provides a technical foundation for realizing integrated infrared imaging systems. Fabricated using a 40-nm CMOS process, the chip occupies an area of 4 000 m × 6 800 m. Experimental results demonstrate that the chip achieved a post-processing non-uniformity of 0.14% and blind pixel rate of 0.19%, meeting the expected requirements for infrared image processing performance.

In this study, a laser-pulse flight-time measurement and data-processing method is designed based on a field-programmable gate array (FPGA) in response to the large-scale, high-precision, highly real-time property and lightweight integration requirements of laser-ranging technology proposed by intelligent unmanned platform. Additionally, the method is devised to address the challenges of complex systems and subpar scalability caused by the dependence of conventional laser-ranging methods on dedicated time-to-digital converters and constant-ratio timing-threshold discrimination circuits. A measurement accuracy of 1 ns is achieved using the high-frequency clock counting of the FPGA and multiphase time interpolation. A pulse-width compensation method is designed to correct the time drift error caused by changes in the laser echo power at different distances. The ranging results of the prototype show that the time accuracy reaches 1 ns and the measuring error is ±0.15 m within the range of 10 m to 5 km, thus satisfying the requirements of large range and high precision simultaneously.

To expand the design freedom of metalenses, realize multifunctional metalenses, and reduce the design difficulty of metasurface devices, an inverse method based on gradient-descent optimization is proposed. Combined with the backpropagation algorithm to solve the gradient information, the polarization-insensitive achromatic metalens design based on a single-layer structure is realized. Using two-photon polymerization three-dimensional (3D) printing technology, the processing technology was investigated, and a metalens with a lateral resolution of 200 nm was processed. Experimental results revealed that the focal length of the processed metalens deviated from the design value by less than 3%. Compared with other studies, this study combines the design flexibility of the inverse design method with the 3D processing advantages of the two-photon polymerization 3D printing method. As such, a new solution for designing and processing high-degree-of-freedom metasurfaces is provided by fully exploiting the advantages of light field control of metasurfaces, thereby promoting the engineering application of metasurfaces.

To meet the demand for high-quality through-silicon via (TSV) machining, in this study, we propose an abrasive-assisted laser-electrochemical hybrid machining method. In this combined method, which leverages laser electrochemical processing and laser-induced cavitation effects, abrasive particles cause high-speed erosion of inner walls, effectively removing surface slag and oxide layers. The effects of abrasive assistance on machining quality are evaluated through a comparative analysis, and the influence of laser power and applied voltage on the processing results is systematically examined. The results show that abrasive-assisted machining substantially improves the quality of laser-electrochemical hybrid processing. At the optimal laser power of 26 W, the TSV inner walls exhibit low oxygen contents (3.27%), low surface roughness (454 nm), and small taper angles (3.26°), which confirm the superior machining performance of the proposed method. At an applied voltage of 30 V, the TSVs exhibit an optimal morphology with smooth sidewalls and a minimal taper angle.

Aluminum antimonide (AlSb) shows broad application prospects in the fields ofhigh-energy particle detectors and solar cells because of its outstanding optoelectronic properties. Here, we report the growth of freestanding AlSb nanowires on Si substrates via molecular beam epitaxy. The epitaxial growth of AlSb nanowires does not require foreign metal catalysts, and the obtained freestanding nanowires have small diameters. In addition, the growth temperature and Ⅴ/Ⅲflux ratios have an obvious effect on the morphology of the nanowires, and the axial growth rate of the nanowires can be improved by increasing the growth temperature. Results of detailed structural investigations confirm that the AlSb nanowires exhibit a zinc-blende phase and are easily susceptible to deliquescence in air. This paper presents a new method for synthesizing AlSb semiconductors on Si substrates.

The flow and heat transfer processes in a wavy double-layer microchannel heat sink were numerically simulated using CFD techniques. The effects of Reynolds number and the upper-layer truncation length ratio on flow and heat transfer performance were analyzed. Based on the field synergy principle and entransy dissipation theory, the flow and heat transfer characteristics of four types of double-layer microchannel heat sinks were comparatively analyzed, revealing the internal mechanisms responsible for enhanced heat transfer. The simulation results show that under laminar flow conditions, the average temperature of the microchannel heat sink gradually decreases as the Reynolds number increases. In contrast, the Nusselt number, surface heat transfer coefficient, and the integrated performance evaluation criterion increase with Reynolds number, while entransy dissipation decreases. At the same Reynolds number, the synergistic angle of heat transfer is smallest when the upper channel length ratio of the wavy-bottom structure is 0.2. Among the configurations studied, the wavy-bottom double-layer microchannel heat sink with an upper channel length ratio of 0.2 exhibits the best overall performance in terms of heat transfer efficiency and flow characteristics.

The hyperspectral unmixing method, based on Transformer, divides a hyperspectral image into small patches of fixed size without overlapping. The divided patches are then fed as inputs into the Transformer encoder and unmixed by a decoder. However, non-overlapping patch division eliminates similar information between neighboring pixels. Furthermore, some local information is lost in the Transformer feedforward layer, which affects the unmixing effect. In this study, we adopt an improved hyperspectral unmixing method with overlapping spectral patch division and deep convolution in the fully connected layer of Transformer. Hyperspectral imaging experiments show that the unmixing effect achieved by the improved method is more pronounced than that of the original method.

To address the disadvantage of the “path-splitting selecting strategy based on a search set under a successive cancellation list” decoding algorithm, a path-splitting strategy based on a reliability function and a pruning strategy relying on an auxiliary path metric (APM) are proposed. Subsequently, a path-splitting selecting strategy based on a reliability function under a successive cancellation list (PSS-RF-SCL) decoding algorithm is proposed. During the decoding of the algorithm, the path metric (PM) values for all paths is calculated before performing path splitting on each information bit. The reliability function value of the bit is calculated using these PM values. Information bits with reliability-function values below its average value (namely, the threshold ) are regarded as bits that require path splitting. This method is used to identify splitting bits and significantly reduces the splitting number. Additionally, paths with APM values above the APM average value (namely, the threshold ) of the correct decoding path are considered unreliable paths. Pruning the unreliable paths significantly reduces the total number of decoding lists. Simulation results show that compared with conventional path-splitting decoding algorithms based on search sets, the proposed PSS-RF-SCL decoding algorithm significantly reduces the decoding complexity without performance loss.

To address the problems of complex algorithm structure and large number of parameters existing in road lane detection, a lane-detection method based on dual-branch feature extraction (DP-RESA) is proposed. This model treats road-lane detection as a semantic segmentation task. First, lane features were extracted using two parallel branches: a semantic branch and a detail branch. The semantic branch uses a lightweight MobileNetV2 model to extract high-level features, while the detail branch uses wide channels and shallow layers to capture low-level details with more spatial details. Furthermore, the model quickly and efficiently fuses features by using high-level features as weights to filter essential semantic information embedded within the low-level features. Finally, the experimental results for the Tusimple dataset show that compared with the baseline, DP-RESA network lane detection accuracy can reach 96.58%. In addition, the number of model parameters is reduced to 5.12 MB, and the single-image inference time is reduced by 11.76 ms, making the model well suited for lane-detection tasks deployed on resource-constrained embedded platforms.

A novel fingertip three-dimensional (3D) force tactile sensor based on fiber Bragg grating is proposed and applied to crack detection. The sensor employs four FBGs combined with a sensitization structure, which optimizes the measurement sensitivity to axial and lateral forces while simplifying the decoupling process. The deformation characteristics under different forces were analyzed using a mechanical simulation, and a 3D force sensing model was constructed. The experimental results show that the sensitivities of the sensor in the X, Y, and Z directions are 1 680.45, 1 441.28, and 125.36 pm/N, respectively, which indicates the ability for 3D detection. In a dynamic experiment, the 3D partial force measured using the sensor is highly consistent with the theoretical value, and the maximum relative error of the combined force is only 4.40%. In the crack detection experiment, it exhibits good crack width detection and force monitoring abilities, and the maximum relative error of the detection width is less than 4%. The sensor is characterized by simple structure, easy decoupling method, and high sensitivity, which is highly applicable to crack detection and the repair tasks of robots in complex environments.

To improve the accuracy of predicting anomalies in optical cable systems based on optical power data, in this study, we employ the simulated annealing-long short-term memory (SA-LSTM) algorithm and predict optical power fluctuations. First, the raw optical power data are normalized. Subsequently, the SA algorithm is used to perform global optimization of the LSTM algorithm’s key parameters, including the number of iterations, hidden layer units, and learning rate, across a wide range. These optimized parameters are used to construct an LSTM model that can predict future trends in optical power. Experimental results demonstrate that the optical power, predicted using the proposed method with optimal parameters, exhibits a root-mean-square error of 2.83 × 10−3 dB, which is only 0.14% of the mean value. The prediction accuracy of the proposed method is 91.7%, 91.0%, 96.4%, and 96.3% higher than those of the autoregressive moving-average radial-basis-function, autoregressive integrated moving-average support-vector-machine, autoregressive moving-average LSTM, and autoregressive integrated moving-average LSTM algorithms, respectively. Thus, the proposed method can predict optical power trends in power cable systems with high accuracy, showing significant potential for engineering applications.

To mitigate the short cycle structure and poor error correction performance of the current quasi-cyclic low-density parity-check (QC-LDPC) codes, a novel construction method, based on the Fibonacci-Lucas sequence (FLS), is proposed for QC-LDPC codes with girth-8. In this construction method, the FLS is used to form an exponent matrix without girth-4, and then, a search algorithm identifies elements that eliminate girth-6 from the exponent matrix, yielding an incremental sequence. The corresponding exponent matrix is constructed to obtain its parity check matrix. Simulation results show that compared with three other QC-LDPC codes with the same code-rate and code-length, the constructed FLS-QC-LDPC code significantly improves the net coding gain at a bit error rate of 1×10−6, outperforming its counterparts in terms of error correction. In addition, the proposed construction method can flexibly select the code length and rate while maintaining a low computational complexity, which is efficient and adaptable.

As the power grid continues to expand and grow more complex each year, the frequency of power grid dispatching operations is increasing. Any misoperation within this process can have serious repercussions. To address these challenges, this study proposes the development of an artificial intelligence model designed to prevent misoperations in power grid dispatching and operation classification. First, we develop a misoperation prevention model based on the combination of convolutional neural networks (CNN) and long short-term memory (LSTM) networks. This CNN-LSTM-based misoperation prevention model for power grid dispatching is designed to explore the deep and temporal characteristics of power grid dispatching operations, achieve more accurate identification of misoperations, and improve the efficiency of power grid dispatching misoperation prevention. Additionally, we propose a refined misoperation classification method that combines the C4.5 decision tree algorithm with Boolean and Bayesian formulas. This approach aims to achieve a more precise classification of misoperations. The simulation results reveal that the proposed algorithm improves the success rate of misoperation prevention by 2.50% to 3.29%, respectively, compared with existing algorithms.

The fixed-size patch division method may cause the local structure of an image to collapse and induce inconsistency in semantics between images during visual Transformer-based hyperspectral unmixing. Therefore, this study proposes a data-driven method that adaptively adjusts the position and scale of patch division according to the input hyperspectral image and effectively captures the local features and spatial positions in the image via a multiscale deformation attention mechanism. In addition, a multilevel feature extraction strategy is used to comprehensively mine the rich features contained in the image. A comparison of experimental results obtained for multiple real datasets shows that the proposed method can accurately estimate the abundance map and endmember spectrum. Moreover, it exhibits excellent generalization ability and stability in different scenarios, especially during the spectral feature processing of complex objects.

Foggy weather often reduces the image quality captured by imaging devices, leading to lower target contrast and blurred contours. While traditional dark channel dehazing algorithms are effective, they tend to produce halo artifacts and excessive smoothing in areas with sharp edges and significant depth variation. To address this, a dark channel dehazing algorithm based on multi-scale edge-aware filtering is proposed. This algorithm introduces edge-aware filtering weights into the guided filtering cost function to adjust the degree of smoothing dynamically. The transmission map is then refined using weighted calculations with multiple edge-aware filters of different scales. Finally, the atmospheric light intensity and the transmission map obtained by the proposed algorithm are substituted into the atmospheric scattering model to recover the haze-free image. Experimental results show that, compared to traditional methods, the proposed algorithm improves image quality in terms of information entropy, average gradient, standard deviation, and color uniformity by 7.60%, 36.19%, 54.25%, and 47.59%, respectively. Overall, the proposed algorithm outperforms traditional methods in dehazing, effectively restores image details, and demonstrates promising application prospects.