187037 research outputs found
Sort by
A facile route to N-doped p-type ZnO via Al co-doping and the fabrication of a p-ZnO/n-GaN light emitting diode
No full textp-ZnO has emerged as a promising alternative to p-GaN, owing to its low-temperature growth process and high hole concentration. Nevertheless, the pursuit of reliable and high p-type conductivity in ZnO is hindered by self-compensation effects, deep acceptor levels, and the inherently low solubility of acceptor dopants. In this study, we successfully fabricated p-type ZnO using a two-step process: initially synthesizing Al-doped ZnO (ZnO:Al) via atomic layer deposition, followed by post-annealing in an ammonia atmosphere at varying temperatures. The resulting Al-N co-doped ZnO (ZnO:Al-N) exhibits a high hole concentration of 2.7 x 1018 cm-3, high hole mobility of 40 cm2 V-1 s-1, and low resistivity of 5.8 x 10-2 52 cm. The optimized p-ZnO was deposited on n-Si to form a heterojunction diode. It demonstrates excellent rectifying behavior, with a forward-to-reverse current ratio of 920 at +/- 2 V, confirming the realization of a high-quality diode. Electrical and optical stability tests further reveal that the p-type conductivity of ZnO:Al-N films remains stable over time, without reversion to ntype conduction. Furthermore, a fully functioning p-ZnO/n-GaN LED device, which incorporated InGaN/GaN quantum wells was fabricated. The electroluminescence spectra revealed stable violet emission centered at 429 nm under a continuous forward current of 24 mA for 430 min, indicating excellent structural and electrical stability of the Al-N co-doped ZnO layer. These results highlight the potential of ZnO:Al-N as a robust p-type material for next-generation III-nitride optoelectronic devices, offering an effective pathway for low-temperature fabrication and improved hole injection efficiency.
Domain-embedded deep learning frameworks for topology optimization: Enhancing structural performance under data scarce environments
No full textRecently, the rapid advancement of deep learning technology has led to the development of numerous topology optimization approaches, significantly reducing computational costs. However, conventional deep learningbased methods inherently suffer from a chronic limitation. They require large-scale data to extract features from the data itself. In particular, in the worst-case scenario where the data is insufficient, these methods may fail to capture the physical characteristics of the structure accurately, potentially leading to physically meaningless and unrealistic results. To solve this problem, this paper proposes an enhanced deep learning model suitable for topology optimization. The main novelty of this study is embedding the feature of topology optimization into a deep learning model. To effectively embed the topology domain, the proposed method introduces three key strategies. Firstly, topology convolutional neural network (CNN) filter layers are incorporated into the neural network model. A CNN is a specialized deep learning architecture designed for grid-structured data such as images, and the topology CNN filter layers are specifically designed to enhance structural connectivity by considering the influence of neighboring elements. Secondly, the pixel-based loss function is augmented with physics-informed loss functions that encapsulate the physical knowledge of topology optimization. Thirdly, a modified output layer is added to prevent zero values in the structure, thereby enhancing numerical stability. Numerical experiments demonstrate that the proposed deep learning approach successfully overcomes the limitations of conventional deep learning methods in data-scarce environments. Furthermore, the results confirm that the proposed method produces designs comparable to the traditional SIMP method.
Health-conscious charging of lithium-ion batteries using Bayesian optimization guided by a semi-empirical aging model
No full textLithium-ion batteries face a trade-off between fast charging and long-term health, with higher charging rates accelerating degradation. Users and applications value this trade-off differently, underscoring the need for personalized charging strategies. Existing optimization approaches either rely on imperfect models (model-based) or suffer from slow convergence (model-free Bayesian optimization (BO)). Moreover, they often overlook real-world usage variability, which critically impacts degradation. We propose a prior knowledge-guided Bayesian optimization (PrBO) framework that integrates a simplified battery model to accelerate convergence, quantify calendar and cycling aging, and improve interpretability. Simulations show that PrBO outperforms both the Taguchi method and standard BO, rapidly identifying faster, capacity-retention-compliant charging strategies across diverse usage scenarios. These results demonstrate PrBO's potential as a practical, health-conscious solution for battery charging optimization.
Nanoplasmonic real-time RT-RPA and CRISPR/Cas12a detection for rapid point-of-care molecular diagnostics
No full textThe rapid and precise detection of nucleic acids is crucial for effective disease diagnosis and management at the point-of-care (POC) level. Here we report a palm-sized plasmonic photothermal platform for real-time on-chip recombinase polymerase amplification (RPA) and CRISPR/Cas12a detection. An ultrathin photothermal nanoplasmonic cavity (PNC) of Au nanoislands (AuNIs) and an aluminum reflector delivers uniform and efficient photothermal heating under white LED illumination. The configuration drives isothermal amplification and CRISPR-mediated cleavage in a single microchamber while a fluorescence microlens array (FMLA) camera records real-time emission. The compact platform detects the SARS-CoV-2 E gene in 25 min at 25.7 copies per cartridge and achieves 100 % concordance with RT-qPCR across 42 clinical samples. This all-in-one platform can offer a robust and cost-effective solution for molecular diagnostics, facilitating scalable and real-time testing of infectious diseases in decentralized POC settings.
A colorimetric strategy for quantifying amino acids using E. coli auxotrophs displaying gold-binding proteins
No full textWe present a cell-based colorimetric assay for amino acid quantification using Escherichia coli auxotrophs engineered to grow exclusively in the presence of their corresponding target amino acids and to display goldbinding proteins (GBPs) on their surface. Upon the addition of gold nanoparticles (AuNPs), the surfaceexpressed GBPs mediate AuNP aggregation, resulting in a distinct color shift from pink to blue. This approach enables the detection of ten essential amino acids through characteristic colorimetric changes and allows for quantitative analysis via absorbance measurements with high specificity, strong linearity (R-2 > 0.96), and low detection limits ranging from 0.43 to 1.04 mu M. The clinical applicability of the assay was demonstrated by accurately detecting elevated amino acid levels associated with phenylketonuria (PKU), homocystinuria, and maple syrup urine disease (MSUD), with excellent analytical precision (coefficients of variation <4 %) and high recovery rates (98-102 %). This technology holds strong potential as a novel platform for amino acid quantification, providing an easily observable colorimetric readout without the need for bulky or specialized instrumentation.
Penetration of biopolymer hydrogels in granular soils for pressurized injection grouting
No full textOne of the unclarified areas in biopolymer-based soil treatment technique is the permeation capabilities of highly concentrated biopolymer hydrogels for grouting applications. This study aims to evaluate the permeation characteristics of xanthan gum (XG) biopolymer-based grouts by identifying key factors influencing penetration efficiency and quantifying their effective viscosity in granular soils. The effects of grout rheology, soil conditions, and injection pressure on the penetrability were investigated through constant-pressure 1-D sand column tests and rheological analyses. Results show that penetration length decreases with lower water-to-biopolymer (W/B) ratio due to increased yield stress, but increases with coarser grain sizes and higher injection pressure. In contrast, silt content of 10-20% significantly impedes grout infiltration. The close agreement between analytical predictions and experimental penetration lengths suggests that flow stoppage is primarily governed by rheological blocking. Furthermore, the study demonstrates that XG-based grout can achieve groutability comparable to conventional grouts under typical permeation grouting conditions. This performance is largely attributed to their strong shear-thinning behavior, which enables a reduction in viscosity by up to four orders of magnitude under flow, facilitating deep penetration through granular media. These findings advance understanding of the relationship between biopolymer grout rheology and permeation dynamics and support the potential of biopolymer hydrogels as a sustainable grouting alternative for permeation grouting applications.
When do ventures break path dependence? Evidence from financial and technological success of serial entrepreneurs
No full textThis study examines ventures' path dependence and path breaking in the context of serial entrepreneurship. While prior entrepreneurial experience is known to cause ventures' path dependence, little is known about how different types of experience, especially financial versus technological, shape it. Specifically, we investigate how the financial and technological success of serial entrepreneurs' prior ventures influences their subsequent ventures' path-breaking behavior singly and jointly. We find that while the financial success of prior ventures facilitates subsequent ventures' path-dependent propensity, technological success leads subsequent ventures to break the path by entering new technology domains different from those of prior ventures. The relationship between prior ventures' financial success and subsequent ventures' path breaking is contingent on prior ventures' technological success. Under the condition of prior ventures' technological success, the negative effect of financial success on path breaking weakens. We corroborate our hypotheses using Crunchbase for entrepreneur and venture data worldwide from 1967 to 2018 and patent data from USPTO. Our findings have important implications for the source of ventures' path-breaking behaviors, highlighting the role of the entrepreneurs' technological success in prior ventures in disentangling their subsequent ventures from path dependence.
In-vivo monitoring of macrophage mitochondrial pH dynamics in zebrafish using an ultrasensitive and water-soluble targeted fluorescent sensor
No full textMitochondrial pH has a key role in cellular metabolism and also may be a sign of pathology. Thus, closely monitoring of minor changes in this value is critical in biological research. However, it proves difficult due to a lack of availability of appropriate probes. A water-soluble fluorescent mitochondrial targeting probe TPP-MpH is reported. The probe has a hemicyanin-based fluorophore, a triphenylphosphonium (TPP) group for mitochondrial targeting and a PEG group for increased water solubility and biocompatibility. The experimental results show that TPP-MpH reflects ultrasensitive responses to different pH values. The experimental results with cell and zebrafish models indicate that TPP-MpH is highly responsive and specific to pH levels, showing excellent stability and minimal toxicity. Confocal fluorescence imaging confirms that TPP-MpH effectively targets mitochondria and accurately monitors their pH fluctuations. To our astonishment, at pH 8.0 and 9.0, TPP-MpH showed expression in specific cells in the gallbladder and liver region of the zebrafish model, these cells are predominantly mitochondria-rich cells that are involved in the uptake of Na+ ions. Hence, these findings open the door to use this probe for further and deep biological studies.
Quantile-based recursive non-additive multi-fidelity emulator with application to seismic response prediction
No full textMulti-fidelity emulation offers an efficient approach to addressing the computational challenges of high-fidelity simulations, which are crucial for predicting complex system behavior in engineering. Traditional models often rely on linear autoregressive structures, which are limited in their ability to capture nonlinear relationships between fidelity levels. The Recursive Non-Additive (RNA) emulator was introduced to overcome these limitations by providing a flexible, nonlinear formulation for modeling interactions across multiple fidelities. However, prior studies of RNA have been restricted to noise-free settings, limiting its applicability to real-world scenarios characterized by stochastic variability. To address this gap, this study integrates an additional Gaussian process (GP) at each fidelity level of the RNA framework to model input-dependent variance, thereby enabling the emulator to account for noisy data and variable noise levels across the input space. In addition, the RNA framework is refined by replacing mean predictions from lower fidelity levels with statistical quantiles, which capture distributional behavior more effectively and improve robustness in representing complex data patterns and uncertainty. The proposed framework is validated through numerical examples, demonstrating improved accuracy and reliability under noisy conditions compared with the original RNA. Furthermore, it is applied to seismic response prediction in structural engineering and evaluated not only against RNA but also against four linear autoregressive multi-fidelity models. The results show improved performance over RNA and competitive, and in some cases superior, performance relative to the linear autoregressive baselines, while maintaining computational efficiency for practical civil engineering applications.
PASSREFINDER-FL: Privacy-preserving credential stuffing risk prediction via graph-based federated learning for representing password reuse between websites
No full textCredential stuffing attacks have caused significant harm to online users who frequently reuse passwords across multiple websites. While prior research has attempted to detect users with reused passwords or identify malicious login attempts, existing methods often compromise usability by restricting password creation or website access, and their reliance on complex account-sharing mechanisms hinders real-world deployment. To address these limitations, we propose PASSREFINDER-FL, a novel framework that predicts credential stuffing risks across web-sites. We introduce the concept of password reuse relations-defined as the likelihood of users reusing passwords between websites-and represent them as edges in a website graph. Using graph neural networks (GNNs), we perform a link prediction task to assess credential reuse risk between sites. Our approach scales to a large number of arbitrary websites by incorporating public website information and linking newly observed websites as nodes in the graph. To preserve user privacy, we extend PASSREFINDER-FL with a federated learning (FL) approach that eliminates the need to share user sensitive information across administrators. Evaluation on a real-world dataset of 360 million breached accounts from 22,378 websites shows that PASSREFINDER-FL achieves an F1-score of 0.9153 in the FL setting. We further validate that our FL-based GNN achieves a 4-11 % performance improvement over other state-of-the-art GNN models through an ablation study. Finally, we demonstrate that the predicted results can be used to quantify password reuse likelihood as actionable risk scores. Our implementation is available at https://github.com/jaehanwork/PassREfinder-FL.