Single-photon avalanche diode fabricated in standard 55†nm bipolar-CMOS-DMOS technology with sub-20†V breakdown voltage.

This paper presents a single-photon avalanche diode (SPAD) in 55 nm bipolar-CMOS-DMOS (BCD) technology. In order to realize a SPAD having sub-20 V breakdown voltage for mobile applications while preventing high tunneling noise, a high-voltage N-well available in BCD is utilized to implement the avalanche multiplication region. The resulting SPAD has a breakdown voltage of 18.4 V while achieving an excellent dark count rate of 4.4 cps/µm2 at the excess bias voltage of 7 V in spite of the advanced technology node. At the same time, the device achieves a high peak photon detection probability (PDP) of 70.1% at 450 nm thanks to the high and uniform E-field. Its PDP values at 850 and 940 nm, wavelengths of interest for 3D ranging applications reach 7.2 and 3.1%, respectively, with the use of deep N-well. The timing jitter of the SPAD, full width at half maximum (FWHM), is 91 ps at 850 nm. It is expected that the presented SPAD enables cost-effective time-of-flight and LiDAR sensors with the advanced standard technology for many mobile applications.

Two-Dimensional Mechanics of Atomically Thin Solids on Water.

Movement of a three-dimensional solid at an air-water interface is strongly influenced by the extrinsic interactions between the solid and the water. The finite thickness and volume of a moving solid causes capillary interactions and water-induced drag. In this Letter, we report the fabrication and dynamical imaging of freely floating MoS2 solids on water, which minimizes such extrinsic effects. For this, we delaminate a synthesized wafer-scale monolayer MoS2 onto a water surface, which shows negligible height difference across water and MoS2. Subsequently patterning by a laser generates arbitrarily shaped MoS2 with negligible in-plane strain. We introduce photoswitchable surfactants to exert a lateral force to floating MoS2 with a spatiotemporal control. Using this platform, we demonstrate a variety of two-dimensional mechanical systems that show reversible shape changes. Our experiment provides a versatile approach for designing and controlling a large array of atomically thin solids on water for intrinsically two-dimensional dynamics and mechanics.

Resist-Free Lithography for Monolayer Transition Metal Dichalcogenides.

Photolithography and electron-beam lithography are the most common methods for making nanoscale devices from semiconductors. While these methods are robust for bulk materials, they disturb the electrical properties of two-dimensional (2D) materials, which are highly sensitive to chemicals used during lithography processes. Here, we report a resist-free lithography method, based on direct laser patterning and resist-free electrode transfer, which avoids unintentional modification to the 2D materials throughout the process. We successfully fabricate large arrays of field-effect transistors using MoS2 and WSe2 monolayers, the performance of which reflects the properties of the pristine materials. Furthermore, using these pristine devices as a reference, we reveal that among the various stages of a conventional lithography process, exposure to a solvent like acetone changes the electrical conductivity of MoS2 the most. This new approach will enable a rational design of reproducible processes for making large-scale integrated circuits based on 2D materials and other surface-sensitive materials.

Noise optimization of single-photon avalanche diodes fabricated in 110 nm CMOS image sensor technology.

This paper presents the effect of shallow trench isolation (STI) on the dark count rate (DCR) and after-pulsing probability (APP) of deep-junction-based single-photon avalanche diodes (SPADs). Two different SPADs were fabricated in 110 nm CMOS image sensor technology, one with STI and the other without STI between its anode and cathode. With TCAD simulations and measurements, we have clearly demonstrated that the SPAD without STI enables a dramatic decrease in DCR by more than three orders of magnitude without suffering from the lateral leakage current between the anode and cathode. By excluding the STI from the device, the proposed SPAD also achieves a negligible APP while the SPAD with STI shows a very high APP of 92%. Thanks to the low-noise performance, the proposed SPAD becomes operable with higher excess bias voltage so that it achieves good photon detection probability, 58.3% at 500 nm and 3% at 940 nm, and timing jitter, 71 ps full width at half maximum at 670 nm, when the reverse bias voltage is 17 V.

Atomically Thin, Optically Isotropic Films with 3D Nanotopography.

Flat optics aims for the on-chip miniaturization of optical systems for high-speed and low-power operation, with integration of thin and lightweight components. Here, we present atomically thin yet optically isotropic films realized by using three-dimensional (3D) topographic reconstruction of anisotropic two-dimensional (2D) films to balance the out-of-plane and in-plane optical responses on the subwavelength scale. We achieve this by conformal growth of monolayer transition metal dichalcogenide (TMD) films on nanodome-structured substrates. The resulting films show an order-of-magnitude increase in the out-of-plane susceptibility for enhanced angular performance, displaying polarization isotropy in the off-axis absorption, as well as improved photoluminescence emission profiles, compared to their flat-film counterparts. We further show that such 3D geometric programming of optical properties is applicable to different TMD materials, offering spectral generalization over for the entire visible range. Our approach presents a powerful platform for advancing the development of atomically thin flat optics with custom-designed light-matter interactions.

A CMOS Optoelectronic Receiver IC with an On-Chip Avalanche Photodiode for Home-Monitoring LiDAR Sensors.

This paper presents an optoelectronic receiver (Rx) IC with an on-chip avalanche photodiode (APD) realized in a 0.18-µm CMOS process for the applications of home-monitoring light detection and ranging (LiDAR) sensors, where the on-chip CMOS P+/N-well APD was implemented to avoid the unwanted signal distortion from bondwires and electro-static discharge (ESD) protection diodes. Various circuit techniques are exploited in this work, such as the feedforward transimpedance amplifier for high gain, and a limiting amplifier with negative impedance compensation for wide bandwidth. Measured results demonstrate 93.4-dBΩ transimpedance gain, 790-MHz bandwidth, 12-pA/√Hz noise current spectral density, 6.74-µApp minimum detectable signal that corresponds to the maximum detection range of 10 m, and 56.5-mW power dissipation from a 1.8-V supply. This optoelectronic Rx IC provides a potential for a low-cost low-power solution in the applications of home-monitoring LiDAR sensors.

A highly tunable and fully biocompatible silk nanoplasmonic optical sensor.

Novel concepts for manipulating plasmonic resonances and the biocompatibility of plasmonic devices offer great potential in versatile applications involving real-time and in vivo monitoring of analytes with high sensitivity in biomedical and biological research. Here we report a biocompatible and highly tunable plasmonic bio/chemical sensor consisting of a natural silk protein and a gold nanostructure. Our silk plasmonic absorber sensor (SPAS) takes advantage of the strong local field enhancement in the metal-insulator-metal resonator in which silk protein is used as an insulating spacer and substrate. The silk insulating spacer has hydrogel properties and therefore exhibits a controllable swelling when exposed to water-alcohol mixtures. We experimentally and numerically show that drastic spectral shifts in reflectance minima arise from the changing physical volume and refractive index of the silk spacer during swelling. Furthermore, we apply this SPAS device as a glucose sensor with a very high sensitivity of 1200 nm/RIU (refractive index units) and high relative intensity change.

A first single-photon avalanche diode fabricated in standard SOI CMOS technology with a full characterization of the device.

This paper reports on the first implementation of a single-photon avalanche diode (SPAD) in standard silicon on insulator (SOI) complementary metal-oxide-semiconductor (CMOS) technology. The SPAD is realized in a circular shape, and it is based on a P(+)/N-well junction along with a P-well guard-ring structure formed by lateral diffusion of two closely spaced N-well regions. The SPAD electric-field profile is analyzed by means of simulation to predict the breakdown voltage and the effectiveness of premature edge breakdown. Measurements confirm these predictions and also provide a complete characterization of the device, including current-voltage characteristics, dark count rate (DCR), photon detection probability (PDP), afterpulsing probability, and photon timing jitter. The SOI CMOS SPAD has a PDP above 25% at 490-nm wavelength and, thanks to built-in optical sensitivity enhancement mechanisms, it is as high as 7.7% at 850-nm wavelength. The DCR is 244 Hz/µm2, and the afterpulsing probability is less than 0.1% for a dead time longer than 200 ns. The SPAD exhibits a timing response without exponential tail and provides a remarkable timing jitter of 65 ps (FWHM). The new device is well suited to operate in backside illumination within complex three-dimensional (3D) integrated circuits, thus contributing to a great improvement of fill factor and jitter uniformity in large arrays.

A fully-integrated 12.5-Gb/s 850-nm CMOS optical receiver based on a spatially-modulated avalanche photodetector.

We present a fully integrated 12.5-Gb/s optical receiver fabricated with standard 0.13-µm complementary metal-oxide-semiconductor (CMOS) technology for 850-nm optical interconnect applications. Our integrated optical receiver includes a newly proposed CMOS-compatible spatially-modulated avalanche photodetector, which provides larger photodetection bandwidth than previously reported CMOS-compatible photodetectors. The receiver also has high-speed CMOS circuits including transimpedance amplifier, DC-balanced buffer, equalizer, and limiting amplifier. With the fabricated optical receiver, detection of 12.5-Gb/s optical data is successfully achieved at 5.8 pJ/bit. Our receiver achieves the highest data rate ever reported for 850-nm integrated CMOS optical receivers.

SNR characteristics of 850-nm OEIC receiver with a silicon avalanche photodetector.

We investigate signal-to-noise ratio (SNR) characteristics of an 850-nm optoelectronic integrated circuit (OEIC) receiver fabricated with standard 0.25-µm SiGe bipolar complementary metal-oxide-semiconductor (BiCMOS) technology. The OEIC receiver is composed of a Si avalanche photodetector (APD) and BiCMOS analog circuits including a transimpedance amplifier with DC-balanced buffer, a tunable equalizer, a limiting amplifier, and an output buffer with 50-Ω loads. We measure APD SNR characteristics dependence on the reverse bias voltage as well as BiCMOS circuit noise characteristics. From these, we determine the SNR characteristics of the entire OEIC receiver, and finally, the results are verified with bit-error rate measurement.

Synephrine inhibits eotaxin-1 expression via the STAT6 signaling pathway.

Citrus contain various flavonoids and alkaloids that have multiple biological activities. It is known that the immature Citrus contains larger amounts of bioactive components, than do the mature plants. Although Citrus flavonoids are well known for their biological activities, Citrus alkaloids have not previously been assessed. In this study, we identified synephrine alkaloids as an active compound from immature Citrus unshiu, and investigated the effect of synephrine on eotaxin-1 expression. Eotaxin-1 is a potent chemoattractant for eosinophils, and a critical mediator, during the development of eosinophilic inflammation. We found that synephrine significantly inhibited IL-4-induced eotaxin-1 expression. This synephrine effect was mediated through the inhibition of STAT6 phosphorylation in JAK/STAT signaling. We also found that eosinophil recruitment induced by eotaxin-1 overexpression was inhibited by synephrine. Taken together, these findings indicate that inhibiting IL-4-induced eotaxin-1 expression by synephrine occurs primarily through the suppression of eosinophil recruitment, which is mediated by inhibiting STAT6 phosphorylation.

Automated detection of large vessel occlusion using deep learning: a pivotal multicenter study and reader performance study.

BACKGROUND: To evaluate the stand-alone efficacy and improvements in diagnostic accuracy of early-career physicians of the artificial intelligence (AI) software to detect large vessel occlusion (LVO) in CT angiography (CTA). METHODS: This multicenter study included 595 ischemic stroke patients from January 2021 to September 2023. Standard references and LVO locations were determined by consensus among three experts. The efficacy of the AI software was benchmarked against standard references, and its impact on the diagnostic accuracy of four residents involved in stroke care was assessed. The area under the receiver operating characteristic curve (AUROC), sensitivity, and specificity of the software and readers with versus without AI assistance were calculated. RESULTS: Among the 595 patients (mean age 68.5±13.4 years, 56% male), 275 (46.2%) had LVO. The median time interval from the last known well time to the CTA was 46.0 hours (IQR 11.8-64.4). For LVO detection, the software demonstrated a sensitivity of 0.858 (95% CI 0.811 to 0.897) and a specificity of 0.969 (95% CI 0.943 to 0.985). In subjects whose symptom onset to imaging was within 24 hours (n=195), the software exhibited an AUROC of 0.973 (95% CI 0.939 to 0.991), a sensitivity of 0.890 (95% CI 0.817 to 0.936), and a specificity of 0.965 (95% CI 0.902 to 0.991). Reading with AI assistance improved sensitivity by 4.0% (2.17 to 5.84%) and AUROC by 0.024 (0.015 to 0.033) (all P<0.001) compared with readings without AI assistance. CONCLUSIONS: The AI software demonstrated a high detection rate for LVO. In addition, the software improved diagnostic accuracy of early-career physicians in detecting LVO, streamlining stroke workflow in the emergency room.

Silicon photonics-wireless interface ICs for micro-/millimeter-wave fiber-wireless networks.

We present two types of Si photonics-wireless interface (PWI) integrated circuits (ICs) realized in standard Si technology. Our PWI ICs convert optical signals into radio-frequency (RF) signals for downlink remote antenna units in fiber-wireless networks. Characterization and modeling of Si avalanche photodetectors (APDs) fabricated in two different Si technologies are carried out and used for PWI IC design. A 5-GHz RF-over-fiber PWI IC composed of APD, preamplifier, and power amplifier (PA) is fabricated in 0.18-µm CMOS technology and its performance is verified by 54-Mb/s wireless local area network data transmission. A 60-GHz baseband-over-fiber PWI IC containing APD, baseband photoreceiver, 60-GHz binary phase-shift keying (BPSK) modulator, and 60-GHz PA is realized in 0.25-µm SiGe BiCMOS technology. Error-free transmission of 1.6-Gb/s BPSK data in 60 GHz with this PWI IC is successfully achieved.

Wafer-scale δ waveguides for integrated two-dimensional photonics.

The efficient, large-scale generation and control of photonic modes guided by van der Waals materials remains as a challenge despite their potential for on-chip photonic circuitry. We report three-atom-thick waveguides-δ waveguides-based on wafer-scale molybdenum disulfide (MoS2) monolayers that can guide visible and near-infrared light over millimeter-scale distances with low loss and an efficient in-coupling. The extreme thinness provides a light-trapping mechanism analogous to a δ-potential well in quantum mechanics and enables the guided waves that are essentially a plane wave freely propagating along the in-plane, but confined along the out-of-plane, direction of the waveguide. We further demonstrate key functionalities essential for two-dimensional photonics, including refraction, focusing, grating, interconnection, and intensity modulation, by integrating thin-film optical components with δ waveguides using microfabricated dielectric, metal, or patterned MoS2.

Improving performance robustness of subject-based brain segmentation software.

Purpose: Artificial intelligence (AI)-based image analysis tools to quantify the brain have become commercialized. However, insufficient data for learning and scanner specificity is a limitation for achieving high quality. In the present study, the performance of personalized brain segmentation software when applied to multicenter data using an AI model trained on data from a single institution was improved. Methods: Preindicators of brain white matter (WM) information from the training dataset were utilized for preprocessing. During learning, data of cognitively normal (CN) individuals from a single center were utilized, and data of CN individuals and Alzheimer disease (AD) patients enrolled in multiple centers were considered the test set. Results: The preprocessing based on the preindicator (dice similarity coefficient [DSC], 0.8567) resulted in a better performance than without (DSC, 0.7921). The standard deviation (SD) of the WM region intensity (DSC, 0.8303) had a more substantial influence on the performance than the average intensity (DSC, 0.6591). When the SD of the test data WM intensity was smaller than the learning data, the performance improved (0.03 increase in lower SD, 0.05 decrease in higher SD). Furthermore, preindicator-based pretreatment increased the correlation of mean cortical thickness of the entire gray matter between Atroscan and FreeSurfer, and data augmentation without preprocessing did not.Both preindicator processing and data augmentation improved the correlation coefficient from 0.7584 to 0.8165. Conclusion: Data augmentation and preindicator-based preprocessing of training data can improve the performance of AI-based brain segmentation software, both increasing the generalizability and stability of brain segmentation software.

Operando Spatial Pressure Mapping Analysis for Prototype Lithium Metal Pouch Cells Under Practical Conditions.

Monitoring and diagnosing the battery status in real-time are of utmost importance for clarifying failure mechanism, improving battery performance, and ensuring safety, particularly under fast charging conditions. Recently, advanced operando techniques have been developed to observe changes in the microstructures of lithium deposits using laboratory-scale cell designs, focusing on understanding the nature of Li metal electrodes. However, the macroscopic spatial inhomogeneity of lithium electroplating/stripping in the prototype pressurized pouch cells has not been measured in real-time under practical conditions. Herein, a new noninvasive operando technique, spatial pressure mapping analysis, is introduced to macroscopically and quantitatively measure spatial pressure changes in a pressurized pouch cell during cycling. Moreover, dynamic spatial changes in the macroscopic morphology of the lithium metal electrode are theoretically visualized by combining operando pressure mapping data with mechanical analyses of cell components. Additionally, under fast charging conditions, the direct correlation between abrupt capacity fading and sudden increases in spatial pressure distribution inhomogeneity is demonstrated through comparative analysis of pouch cells under various external pressures, electrolyte species, and electrolyte weight to cell capacity (e/c) ratios. This operando technique provides insights for assessing the current battery status and understanding the complex origin of cell degradation behavior in pressurized pouch cells.

Region Segmentation of Whole-Slide Images for Analyzing Histological Differentiation of Prostate Adenocarcinoma Using Ensemble EfficientNetB2 U-Net with Transfer Learning Mechanism.

Recent advances in computer-aided detection via deep learning (DL) now allow for prostate cancer to be detected automatically and recognized with extremely high accuracy, much like other medical diagnoses and prognoses. However, researchers are still limited by the Gleason scoring system. The histopathological analysis involved in assigning the appropriate score is a rigorous, time-consuming manual process that is constrained by the quality of the material and the pathologist's level of expertise. In this research, we implemented a DL model using transfer learning on a set of histopathological images to segment cancerous and noncancerous areas in whole-slide images (WSIs). In this approach, the proposed Ensemble U-net model was applied for the segmentation of stroma, cancerous, and benign areas. The WSI dataset of prostate cancer was collected from the Kaggle repository, which is publicly available online. A total of 1000 WSIs were used for region segmentation. From this, 8100 patch images were used for training, and 900 for testing. The proposed model demonstrated an average dice coefficient (DC), intersection over union (IoU), and Hausdorff distance of 0.891, 0.811, and 15.9, respectively, on the test set, with corresponding masks of patch images. The manipulation of the proposed segmentation model improves the ability of the pathologist to predict disease outcomes, thus enhancing treatment efficacy by isolating the cancerous regions in WSIs.

An integrated 12.5-Gb/s optoelectronic receiver with a silicon avalanche photodetector in standard SiGe BiCMOS technology.

An optoelectronic integrated circuit (OEIC) receiver is realized with standard 0.25-µm SiGe BiCMOS technology for 850-nm optical interconnect applications. The OEIC receiver consists of a Si avalanche photodetector, a transimpedance amplifier with a DC-balanced buffer, a tunable equalizer, and a limiting amplifier. The fabricated OEIC receiver successfully detects 12.5-Gb/s 2(31)-1 pseudorandom bit sequence optical data with the bit-error rate less than 10(-12) at incident optical power of -7 dBm. The OEIC core has 1000 µm x 280 µm chip area, and consumes 59 mW from 2.5-V supply. To the best of our knowledge, this OEIC receiver achieves the highest data rate with the smallest sensitivity as well as the best power efficiency among integrated OEIC receivers fabricated with standard Si technology.

Robotic four-dimensional pixel assembly of van der Waals solids.

Van der Waals (vdW) solids can be engineered with atomically precise vertical composition through the assembly of layered two-dimensional materials1,2. However, the artisanal assembly of structures from micromechanically exfoliated flakes3,4 is not compatible with scalable and rapid manufacturing. Further engineering of vdW solids requires precisely designed and controlled composition over all three spatial dimensions and interlayer rotation. Here, we report a robotic four-dimensional pixel assembly method for manufacturing vdW solids with unprecedented speed, deliberate design, large area and angle control. We used the robotic assembly of prepatterned 'pixels' made from atomically thin two-dimensional components. Wafer-scale two-dimensional material films were grown, patterned through a clean, contact-free process and assembled using engineered adhesive stamps actuated by a high-vacuum robot. We fabricated vdW solids with up to 80 individual layers, consisting of 100 × 100 µm2 areas with predesigned patterned shapes, laterally/vertically programmed composition and controlled interlayer angle. This enabled efficient optical spectroscopic assays of the vdW solids, revealing new excitonic and absorbance layer dependencies in MoS2. Furthermore, we fabricated twisted N-layer assemblies, where we observed atomic reconstruction of twisted four-layer WS2 at high interlayer twist angles of ≥4°. Our method enables the rapid manufacturing of atomically resolved quantum materials, which could help realize the full potential of vdW heterostructures as a platform for novel physics2,5,6 and advanced electronic technologies7,8.

Diagnostic performance of artificial intelligence model for pneumonia from chest radiography.

OBJECTIVE: The chest X-ray (CXR) is the most readily available and common imaging modality for the assessment of pneumonia. However, detecting pneumonia from chest radiography is a challenging task, even for experienced radiologists. An artificial intelligence (AI) model might help to diagnose pneumonia from CXR more quickly and accurately. We aim to develop an AI model for pneumonia from CXR images and to evaluate diagnostic performance with external dataset. METHODS: To train the pneumonia model, a total of 157,016 CXR images from the National Institutes of Health (NIH) and the Korean National Tuberculosis Association (KNTA) were used (normal vs. pneumonia = 120,722 vs.36,294). An ensemble model of two neural networks with DenseNet classifies each CXR image into pneumonia or not. To test the accuracy of the models, a separate external dataset of pneumonia CXR images (n = 212) from a tertiary university hospital (Gachon University Gil Medical Center GUGMC, Incheon, South Korea) was used; the diagnosis of pneumonia was based on both the chest CT findings and clinical information, and the performance evaluated using the area under the receiver operating characteristic curve (AUC). Moreover, we tested the change of the AI probability score for pneumonia using the follow-up CXR images (7 days after the diagnosis of pneumonia, n = 100). RESULTS: When the probability scores of the models that have a threshold of 0.5 for pneumonia, two models (models 1 and 4) having different pre-processing parameters on the histogram equalization distribution showed best AUC performances of 0.973 and 0.960, respectively. As expected, the ensemble model of these two models performed better than each of the classification models with 0.983 AUC. Furthermore, the AI probability score change for pneumonia showed a significant difference between improved cases and aggravated cases (Δ = -0.06 ± 0.14 vs. 0.06 ± 0.09, for 85 improved cases and 15 aggravated cases, respectively, P = 0.001) for CXR taken as a 7-day follow-up. CONCLUSIONS: The ensemble model combined two different classification models for pneumonia that performed at 0.983 AUC for an external test dataset from a completely different data source. Furthermore, AI probability scores showed significant changes between cases of different clinical prognosis, which suggest the possibility of increased efficiency and performance of the CXR reading at the diagnosis and follow-up evaluation for pneumonia.