Search Results (16)

This paper takes a look at training a generator model based on a limited dataset that can fit the distribution of the original dataset, improving the reconstruction ability of wafer datasets. High-fidelity wafer defect image generation remains challenging due to limited real data and poor physical authenticity of existing methods. We propose an enhanced StyleGANv3 framework with two key innovations: (1) a Heterogeneous Kernel Fusion Unit (HKFU) enabling multi-scale defect feature refinement via spatiotemporal attention and dynamic gating; (2) a Dynamic Adaptive Attention Module (DAAM) adaptively boosting discriminator sensitivity. Experiments on Mixtype-WM38 and MIR-WM811K datasets demonstrate state-of-the-art performance, achieving FID scores of 25.20 and 28.70 alongside SDS values of 36.00 and 35.45. The proposed method in this article helps alleviate the problem of limited datasets and makes an important contribution to data preparation for downstream classification and detection tasks. Full article

Silicon (Si) and silicon carbide (SiC) are second- and third-generation semiconductor materials with excellent properties that are particularly suitable for applications in scenarios such as high temperature, high voltage, and high frequency. Si/SiC wafers face warpage and bending problems during production, which can seriously affect subsequent processing. Fast, accurate, and comprehensive detection of thickness, thickness variation, and flatness (including bow and warpage) of SiC and Si wafers is an industry-recognized challenge. Frequency scanning interferometry (FSI) can synchronize the upper and lower surfaces and thickness information of transparent parallel thin wafers, but it is still affected by multiple interfacial harmonic reflections, reflectivity asymmetry, and phase modulation uncertainty when measuring SiC thin wafers, which leads to thickness calculation errors and face reconstruction deviations. To this end, this paper proposes a high-precision facet reconstruction method for SiC/Si structures, which combines harmonic spectral domain decomposition, refractive index gradient constraints, and partitioning optimization strategy, and introduces interferometric signal “oversampling” and weighted fusion of multiple sets of data to effectively suppress higher-order harmonic interferences, and to enhance the accuracy of phase resolution. The multi-layer iterative optimization model further enhances the measurement accuracy and robustness of the system. The flatness measurement system constructed based on this method can realize the simultaneous acquisition of three-dimensional top and bottom surfaces on 6-inch Si/SiC wafers, and accurately reconstruct the key parameters, such as flatness, warpage, and thickness variation (TTV). A comparison with the Corning Tropel FlatMaster commercial system shows that this method has high consistency and good applicability. Full article

In semiconductor manufacturing, the final testing phase is critical for ensuring chip quality and operational efficiency. Accurate yield prediction at this stage optimizes testing workflows, boosts production efficiency, and enhances quality control. However, existing research primarily focuses on wafer-level yield prediction, leaving the unique challenges of final testing—such as test condition variability and complex failure patterns—insufficiently addressed. This is especially critical for Radio Frequency Front-End (RFFE) chips, where high precision is essential, highlighting the need for a specialized prediction approach. In our study, a rigorous RF correlation parameter selection process was applied, leveraging metrics such as Spearman’s correlation coefficient and variance inflation factors to identify key RF-related features, such as multiple frequency-point PAE measurements and other critical electrical parameters, that directly influence final test yield. To overcome the limitations of traditional methods, this study proposes a multistrategy dynamic weighted fusion model for yield prediction. The proposed approach combines convolutional neural networks (CNNs) and long short-term memory (LSTM) networks with sliding window averaging to capture both local features and long-term dependencies in RFFE test data, while employing a learnable weighting mechanism to dynamically fuse outputs from multiple submodels for enhanced prediction accuracy. It further incorporates incremental training to adapt to shifting production conditions and utilizes principal component analysis (PCA) in data preprocessing to reduce dimensionality and address multicollinearity. Evaluated on a dataset of over 24 million RFFE chips, the proposed model achieved a Mean Absolute Error (MAE) below 0.84% and a Root Mean Square Error (RMSE) of 1.24%, outperforming single models by reducing MAE and RMSE by 7.69% and 13.29%, respectively. These results demonstrate the high accuracy and adaptability of the fusion model in predicting semiconductor final test yield. Full article

Microelectronics and electronic products are integral to our increasingly connected world, facing constant challenges in terms of quality, security, and provenance. As technology advances and becomes more complex, the demand for automated solutions to verify the quality and origin of components assembled on printed circuit boards (PCBs) is skyrocketing. This paper proposes an innovative approach to detecting and classifying microelectronic components with impressive accuracy and reliability, paving the way for a more efficient and safer electronics industry. Our approach introduces significant advancements by integrating optical and X-ray imaging, overcoming the limitations of traditional methods that rely on a single imaging modality. This method uses a novel data fusion technique that enhances feature visibility and detectability across various component types, crucial for densely packed PCBs. By leveraging the WaferCaps capsule network, our system improves spatial hierarchy and dynamic routing capabilities, leading to robust and accurate classifications. We employ decision-level fusion across multiple classifiers trained on different representations—optical, X-ray, and fused images—enhancing accuracy by synergistically combining their predictive strengths. This comprehensive method directly addresses challenges surrounding concurrency, reliability, availability, and resolution in component identification. Through extensive experiments, we demonstrate that our approach not only significantly improves classification metrics but also enhances the learning and identification processes of PCB components, achieving a remarkable total accuracy of 95.2%. Our findings offer a substantial contribution to the ongoing development of reliable and accurate automatic inspection solutions in the electronics manufacturing sector. Full article

With the rapid development of intelligent manufacturing and electronic information technology, integrated circuits play a vital role in high-end chips. The semiconductor chip manufacturing process requires precise operation and strict control to ensure chip quality. The traditional manual visual inspection method has a high workforce cost and intense subjectivity and is accompanied by a high level of misdetection and leakage. Computer vision-based wafer defect detection technology is gaining popularity in the industry. However, previous methods still find it challenging to meet the production requirements regarding accuracy. To solve the problem, we propose a defect detection network based on a coding and decoding structure, Dense Skip-Connection U-Net (DSCU-Net), which optimizes the skip connection between the encoder and decoder and enhances the profound fusion of high-level semantics and low-level semantics to improve accuracy. To verify the effectiveness of DSCU-Net, we validate it in actual microelectromechanical systems (MEMS) data, and the results show that DSCU-Net reaches an optimal level. Therefore, the DSCU-Net proposed in this paper effectively solves the defect detection problem in semiconductor chip manufacturing. This method reduces workforce cost and subjectivity interference and improves inspection efficiency and accuracy. It will help to promote further development in the field of intelligent manufacturing and electronic information technology. Full article

Wafer characters are used to record the transfer of important information in industrial production and inspection. Wafer character recognition is usually used in the traditional template matching method. However, the accuracy and robustness of the template matching method for detecting complex images are low, which affects production efficiency. An improved model based on YOLO v7-Tiny is proposed for wafer character recognition in complex backgrounds to enhance detection accuracy. In order to improve the robustness of the detection system, the images required for model training and testing are augmented by brightness, rotation, blurring, and cropping. Several improvements were adopted in the improved YOLO model, including an optimized spatial channel attention model (CBAM-L) for better feature extraction capability, improved neck structure based on BiFPN to enhance the feature fusion capability, and the addition of angle parameter to adapt to tilted character detection. The experimental results showed that the model had a value of 99.44% for mAP@0.5 and an F1 score of 0.97. In addition, the proposed model with very few parameters was suitable for embedded industrial devices with small memory, which was crucial for reducing the hardware cost. The results showed that the comprehensive performance of the improved model was better than several existing state-of-the-art detection models. Full article

A detailed experimental analysis of the impact of device topology on the performance of 1550 nm VCSELs with an active region based on thin InGaAs/InAlGaAs quantum wells and a composite InAlGaAs buried tunnel junction is presented. The high-speed performance of the lasers with L-type device topology (with the largest double-mesa sizes) is mainly limited by electrical parasitics showing noticeable damping of the relaxation oscillations. For the S-type device topology (with the smallest double-mesa sizes), the decrease in the parasitic capacitance of the reverse-biased p⁺n-junction region outside the buried tunnel junction region allowed to raise the parasitic cutoff frequency up to 13–14 GHz. The key mechanism limiting the high-speed performance of such devices is thus the damping of the relaxation oscillations. VCSELs with S-type device topology demonstrate more than 13 GHz modulation bandwidth and up to 37 Gbps nonreturn-to-zero data transmission under back-to-back conditions at 20 °C. Full article

Single-mode long-wavelength (LW) vertical-cavity surface-emitting lasers (VCSELs) present an inexpensive alternative to DFB-lasers for data communication in next-generation giga data centers, where optical links with large transmission distances are required. Narrow wavelength-division multiplexing systems demand large bit rates and single longitudinal and transverse modes. Spatial division multiplexing transmission through multicore fibers using LW VCSELs is enabling still larger-scale data center networks. This review discusses the requirements for achieving high-speed modulation, as well as the state-of-the-art. The hybrid short-cavity concept allows for the realization of f_3dB frequencies of 17 GHz and 22 GHz for 1300 nm and 1550 nm range VCSELs, respectively. Wafer-fusion (WF) concepts allow the realization of long-time reliable LW VCSELs with a record single-mode output power of more than 6 mW, 13 GHz 3 dB cut-off frequency, and data rates of 37 Gbit/s for non-return-to-zero (NRZ) modulation at 1550 nm. Full article

Artificial membrane systems can serve as models to investigate molecular mechanisms of different cellular processes, including transport, pore formation, and viral fusion. However, the current, such as SUVs, GUVs, and the supported lipid bilayers suffer from issues, namely high curvature, heterogeneity, and surface artefacts, respectively. Freestanding membranes provide a facile solution to these issues, but current systems developed by various groups use silicon or aluminum oxide wafers for fabrication that involves access to a dedicated nanolithography facility and high cost while conferring poor membrane stability. Here, we report the development, characterization and applications of an easy-to-fabricate suspended lipid bilayer (SULB) membrane platform leveraging commercial track-etched porous filters (PCTE) with defined microwell size. Our SULB system offers a platform to study the lipid composition-dependent structural and functional properties of membranes with exceptional stability. With dye entrapped in PCTE microwells by SULB, we show that sphingomyelin significantly augments the activity of pore-forming toxin, Cytolysin A (ClyA) and the pore formation induces lipid exchange between the bilayer leaflets. Further, we demonstrate high efficiency and rapid kinetics of membrane fusion by dengue virus in our SULB platform. Our suspended bilayer membrane mimetic offers a novel platform to investigate a large class of biomembrane interactions and processes. Full article

Predicting the wafer material removal rate (MRR) is an important step in semiconductor manufacturing for total quality control. This work proposes a deep learning model called a fusion network to predict the MRR, in which we consider separating features into shallow and deep features and use the characteristics of deep learning to perform a fusion of these two kinds of features. In the proposed model, the deep features go through a sequence of nonlinear transformations and the goal is to learn the complex interactions among the features to obtain the deep feature embeddings. Additionally, the proposed method is flexible and can incorporate domain knowledge into the model by encoding the knowledge as shallow features. Once the learning of deep features is completed, the proposed model uses the shallow features and the learned deep feature embeddings to obtain new features for the subsequent layers. This work performs experiments on a dataset from the 2016 Prognostics and Health Management Data Challenge. The experimental results show that the proposed model outperforms the competition winner and three ensemble learning methods. The proposed method is a single model, whereas the comparison methods are ensemble models. Besides the experimental results, we conduct extensive experiments to analyze the proposed method. Full article

A bonding process was developed for glass-to-glass fusion bonding using Borofloat 33 wafers, resulting in high bonding yield and high flexural strength. The Borofloat 33 wafers went through a two-step process with a pre-bond and high-temperature bond in a furnace. The pre-bond process included surface activation bonding using O₂ plasma and N₂ microwave (MW) radical activation, where the glass wafers were brought into contact in a vacuum environment in an EVG 501 Wafer Bonder. The optimal hold time in the EVG 501 Wafer bonder was investigated and concluded to be a 3 h hold time. The bonding parameters in the furnace were investigated for hold time, applied force, and high bonding temperature. It was concluded that the optimal parameters for glass-to-glass Borofloat 33 wafer bonding were at 550 °C with a hold time of 1 h with 550 N of applied force. Full article

A high-aspect-ratio three-dimensionally (3D) stacked comb structure for micromirror application is demonstrated by wafer bonding technology in CMOS-compatible processes in this work. A vertically stacked comb structure is designed to circumvent any misalignment issues that could arise from multiple wafer bonding. These out-of-plane comb drives are used for the bias actuation to achieve a larger tilt angle for micromirrors. The high-aspect-ratio mechanical structure is realized by the deep reactive ion etching of silicon, and the notching effect in silicon-on-insulator (SOI) wafers is minimized. The low-temperature bonding of two patterned wafers is achieved with fusion bonding, and a high bond strength up to 2.5 J/m² is obtained, which sustains subsequent processing steps. Furthermore, the dependency of resonant frequency on device dimensions is studied systematically, which provides useful guidelines for future design and application. A finalized device fabricated here was also tested to have a resonant frequency of 17.57 kHz and a tilt angle of 70° under an AC bias voltage of 2 V. Full article

The advantages of the capacitive micromachined ultrasound transducer (CMUT) technology have provided revolutionary advances in ultrasound imaging. Extensive research on CMUT devices for high-frequency medical imaging applications has been conducted because of strong demands and fabrication realization by using standard silicon IC fabrication technology. However, CMUT devices for low-frequency underwater imaging applications have been rarely researched because it is difficult to fabricate thick membrane structures through depositing processes using standard IC fabrication technology due to stress-related problems. To address this shortcoming, in this paper, a CMUT device with a 2.83-μm thick silicon membrane is proposed and fabricated. The CMUT device is fabricated using silicon fusion wafer-bonding technology. A 5-μm thick Parylene-C is conformally deposited on the device for immersion measurement. The results show that the fabricated CMUT can transmit an ultrasound wave, receive an ultrasound wave, and have pulse-echo measurement capability. The ability of the device to emit and receive ultrasonic waves increases with the bias voltage but does not depend on the voltage polarity. The results demonstrate the viability of the fabricated CMUT in low-frequency applications from the perspectives of the device structure, fabrication, and characterization. This study presents the potential of the CMUT for underwater ultrasound imaging applications. Full article

The fabrication of high purity copper using additive manufacturing has proven difficult because of oxidation of the powder feedstock. Here, we present work on the hydrogen heat treatment of copper powders for electron beam powder bed fusion (EB-PBF), in order to enable the fabrication of high purity copper components for applications such as accelerator components and vacuum electronic devices. Copper powder with varying initial oxygen contents were hydrogen heat-treated and characterized for their chemistry, morphology, and microstructure. Higher initial oxygen content powders were found to not only reduce surface oxides, but also reduce oxides along the grain boundaries and form trapped H₂O vapor inside the particles. The trapped H₂O vapor was verified by thermogravimetric analysis (TGA) and residual gas analysis (RGA) while melting. The mechanism of the H₂O vapor escaping the particles was determined by in-situ SEM heated stage experiments, where the particles were observed to crack along the grain boundaries. To determine the effect of the EB-PBF processing on the H₂O vapor, the thermal simulation and the validation of single melt track width wafers were conducted along with melting single layer discs for chemistry analysis. A high speed video of the EB-PBF melting was performed in order to determine the effect of the trapped H₂O vapor on the melt pool. Finally, solid samples were fabricated from hydrogen-treated copper powder, where the final oxygen content measured ~50 wt. ppm, with a minimal residue hydrogen content, indicating the complete removal of trapped H₂O vapor from the solid parts. Full article

With the aim of constructing an S-layer fusion protein that combines both excellent self-assembly and specific ligand i.e., biotin binding ability, streptavidin (aa 16-133) was fused to the S-layer protein of Sporosarcina ureae ATCC 13881 (SslA) devoid of its N-terminal 341 and C-terminal 172 amino acids. The genetically engineered chimeric protein could be successfully produced in E. coli, isolated, and purified via Ni affinity chromatography. In vitro recrystallisation experiments performed with the purified chimeric protein in solution and on a silicon wafer have demonstrated that fusion of the streptavidin domain does not interfere with the self-assembling properties of the S-layer part. The chimeric protein self-assembled into multilayers. More importantly, the streptavidin domain retained its full biotin-binding ability, a fact evidenced by experiments in which biotinylated quantum dots were coupled to the fusion protein monomers and adsorbed onto the in vitro recrystallised fusion protein template. In this way, this S-layer fusion protein can serve as a functional template for the controlled immobilization of biotinylated and biologically active molecules. Full article