Explore the vast range of titles available.

In the post-Moore era, carbon nanotube field-effect transistors (CNTFETs) are a promising alternative to complementary metal-oxide-semiconductor (CMOS) technology at and below the 5 nm node. Compact models bridge circuit design and device physics, yet the electrostatic discharge (ESD) behavior of CNTFETs remains insufficiently captured. Focusing on the local bottom-gate (LBG) CNTFET structure, which offers enhanced gate control due to its bottom-gate configuration, this paper investigates three dominant ESD-triggering mechanisms—thermionic current, tunneling leakage current, and thermal failure breakdown. Then, a hybrid compact–behavioral ESD model for CNTFETs is established. After theoretical derivation and comparison with test results, the model parameters are optimized through fitting. The simulation results exhibit excellent agreement with CNTFET measurements, particularly capturing the Human Body Model (HBM) pre-charge threshold phenomenon at 72 V and accurately predicting the subsequent voltage collapse behavior. This validates the accuracy and effectiveness of the model, laying a theoretical and experimental foundation for further construction of carbon-based standard-cell and I/O libraries.

An Analog-to-Digital Converter (ADC) is an indispensable part of image sensor systems. This paper presents a silicon-based 13-bit 100 kS/s two-step single-slope analog-to-digital converter (TS-SS ADC) for infrared image sensors with a frame rate of 100 Hz. For the charge leakage and offset voltage issues inherent in conventional TS-SS ADC, a four-terminal comparator was employed to resolve the fine ramp voltage offset caused by charge redistribution in storage and parasitic capacitors. In addition, a current-steering digital-to-analog converter (DAC) was adopted to calibrate the voltage reference of the dynamic comparator and mitigate differential nonlinearity (DNL)/integral nonlinearity (INL). To eliminate quantization dead zones, a 1-bit redundancy was incorporated into the fine quantization circuit. Finally, the quantization scheme consisted of 7-bit coarse quantization followed by 7-bit fine quantization. The ADC was implemented using an SMIC 55 nm processSemiconductor Manufacturing International Corporation, Shanghai, China. The post-simulation results show that when the power supply is 3.3 V, the ADC achieves a quantization range of 1.3 V–3 V. Operating at a 100 kS/s sampling rate, the proposed ADC exhibits an effective number of bits (ENOBs) of 11.86, a spurious-free dynamic range (SFDR) of 97.45 dB, and a signal-to-noise-and-distortion ratio (SNDR) of 73.13 dB. The power consumption of the ADC was 22.18 mW.

For the application of a high-performance shortwave infrared (SWIR) detector, a fully integrated analog front-end (AFE) circuit is proposed in this paper, which includes a readout integrated circuit (ROIC) and a 12-bit/100 kHz two-step single-slope analog-to-digital converter (TS-SS ADC). The ROIC adopts a direct injection (DI) structure with a pixel size of only 10 µm × 10 µm. The column processing circuit uses a passive correlated double-sampling (CDS) circuit to reduce noise and improve dynamic range. The comparator of four inputs in the ADC solves the problem of linearity reduction caused by charge redistribution during coarse quantization. In addition, the current steering digital-to-analog converter (DAC) is used to compensate for the non-ideal characteristics of the switch, which effectively optimizes the differential nonlinearity (DNL) and integral nonlinearity (INL). The AFE is implemented using SMIC 180 nm 1P6M technology. The post-simulation results show that at a power supply voltage of 3.3 V, the AFE has a frame rate of 100 Hz and a full well capacity (FWC) of 2.8 Me−. The linearity can reach 99.59%, and the equivalent output noise is 243 µV. The dynamic range is 73.8 dB. Meanwhile, the signal-to-noise distortion ratio (SNDR) and effective number of bits (ENOB) are 68.38 dB and 11.06 bits, respectively.

As a critical component of computer systems, static random access memory (SRAM) plays a significant role in the fields of high-performance computing, artificial intelligence, and the Internet of Things. However, the conventional silicon-based SRAM is facing problems such as high power consumption, integration limitations, and the need for performance improvement. For the power consumption challenge of SRAM, a novel read/write-decoupled SRAM cell is proposed in this paper. When SRAM is in a standby state, utilization of this cell can effectively reduce the leakage current, thereby reducing the static power consumption. In addition, a novel carbon–silicon heterogeneous-integrated SRAM technology is proposed in this paper. The carbon–silicon heterogeneous-integrated SRAM technology combines the high mobility, low power consumption, and low temperature process compatibility of carbon nanotubes (CNTs) with mature silicon-based fabrication processes, enabling significant optimization of access time and power consumption. The SRAM was implemented by a silicon-based 55 nm process and a carbon-based 500 nm process. The simulation results showed that when the power supply was 1.2 V, the access time of SRAM with the silicon 55 nm process was 886 ps. The static and dynamic power consumption was 91.51 nW and 1.48 mW, respectively. The SRAM with carbon–silicon heterogeneous-integrated technology achieved an access time of 872 ps, a static power consumption of 62.54 nW, and a dynamic power consumption of 1.07 mW, representing reductions of 1.58%, 31.66%, and 27.70%, respectively.

The decentralized clinical trial (DCT) approach may potentially improve patient accrual and diversity and reduce logistical burden; however, its implementation in China is still in its infancy due to limited practical experience and guidance. We utilized stakeholder engagement research to simulate DCT processes and investigate their feasibility and applicability in a local context. This study, conducted between October 2023 and June 2024, engaged stakeholders to simulate the roles of the investigator, study nurse/pharmacist, and patient in a DCT, and explored feasibility, challenges, and procedures locally. Using a mixed‐method approach including focus group interviews, heuristic evaluation, usability testing, surveys, and one‐on‐one interviews, four DCT elements (eConsent, direct‐from‐patient sample collection, televisits, and direct‐to‐patient investigational medicine delivery) were simulated at 11 Chinese investigational sites with 36 researchers. Pain points, stakeholder requirements, and key challenges were identified; standardized procedures and a user journey map were developed to provide a visual representation of the DCT process, encouraging broader adoption. The researchers reported high satisfaction across all DCT elements. Key challenges included patients' DCT system process ease of use, lack of thoroughness or attention remotely, integrating online (system) and offline (hospital/home) visits, patient risk management, maintaining medication compliance remotely, and DCT system compatibility, among others. Standard procedures were developed for each DCT element, addressing issues like identity verification and data privacy based on user perspectives. This study provided insights on implementing DCTs in China, identifying areas that need optimization to meet local regulatory and stakeholders' requirements. The standardized framework may be referenced when adapting future DCT elements.

N,N‐dimethylacetamide (DMA) is an organic solvent, used in busulphan, iv‐formulation, (Busulfex). DMA is metabolized primarily to N‐methylacetamide (MMA) via CYP2E1. In the present study, we investigated the pharmacokinetics and toxicity of DMA and MMA in patients, mice and cell lines. In pediatric patients (6 months to 18 years) undergoing hematopoietic stem cell transplantation, the pharmacokinetics of DMA and MMA were followed during the 4‐days Busulfex conditioning (3.2–4 mg/kg, b.i.d). No accumulation of DMA was found; however, a significant increase in clearance and a significant decrease (p < 0.005) in half‐lives by the end of treatment were observed which might indicate CYP2E1 autoinduction. Furthermore, continuous increases in plasma concentration of MMA were observed during treatment and a terminal half‐life of 15.2 ± 1.7 h was detected. Moreover, ALT was significantly (p = 0.04) increased in > 61% of the patients after conditioning. Additionally, mice were treated with either dimethyl sulfoxide (DMSO), DMA, busulphan in DMSO, Busulfex or saline for 4 days. DMA‐treated mice showed elevated ALT and AST values. Interestingly, Busulfex administration did not alter mice liver enzymes. Busulphan in DMA showed higher cytotoxicity compared to busulphan in DMSO in HepG‐2, Huh‐7 and HL‐60 cells. The combination Bu‐DMA‐MMA exhibited increased cytotoxicity in a concentration‐dependent manner. In conclusion, Busulfex administration to pediatric patients resulted in an accumulation of MMA during the 4‐days treatment. Busulfex is less toxic compared to Bu dissolved in DMA. MMA addition to Bu in DMA showed higher cytotoxicity. Therefore, MMA with a relatively long half‐life may induce hepatotoxicity and/or interaction with subsequently administered drugs; thus further investigations are urgently warranted. Study Highlights What is the current knowledge on the topic? Busulfex is an intravenous formulation of busulphan that was introduced in order to decrease the inter‐individual variability in pharmacokinetics compared to oral busulphan. In this formula, the drug is dissolved in polyethylene glycol 400 and N,N‐Dimethylacetamide (DMA). What question did this study address? To investigate the pharmacokinetics of DMA and its primary metabolite N‐methylacetamide (MMA) in pediatric patients undergoing stem cell transplantation. Is there any potential toxicity that can be caused by DMA or its metabolites? What does this study add to our knowledge? Busulfex administration elevated ALT levels in the majority of pediatric patients. Busulfex administration causes MMA (but not DMA) accumulation in patients' plasma. Continuous increases in DMA clearance during 4 days of treatment might indicate an auto‐induction of its metabolism. MMA showed higher cytotoxicity on several cell lines. Accordingly, MMA with a relatively long half‐life may induce hepatotoxicity in patients, especially for poor metabolizers due to polymorphisms (mainly CYP2E1 for DMA metabolism to MMA). The high levels of MMA in combination with its long half‐life might cause undesirable long‐term liver toxicity and/or interaction with subsequently administered drugs. How might this change clinical pharmacology or translational science? The effect of CYP2E1 polymorphisms on DMA metabolism and any potential drug interactions should be further investigated. Patients who are at high risk for hepatotoxicity may need to be monitored closely for DMA/MMA pharmacokinetics and plasma concentration. A prophylactic treatment strategy may be considered for those patients.

Pathogenic blue ear disease caused by porcine reproductive and respiratory syndrome virus (PRRSV) bring severe loss to breeding industry due to high infectivity and mortality. L. plantarum serves as the probiotic host strain, known for its beneficial properties in the gut microbiota. E. coli is used as a cloning host for the initial genetic engineering steps, facilitating the construction and amplification of the desired genetic constructs. In this study, using synthetic biology technology, we constructed engineered probiotics which could adhere and display nanobody on the surface to neutralize virus. Firstly, we screen an optimal nanobody to effectively bind with PRRSV by building library, expression and purification. Then, the integration of adhesion protein and nanobody into the genome of probiotics significantly improved its adhesion to IPEC-J2 cells. In addition, this engineered probiotic is almost non-toxic to cells with good safety, which can be used as a daily probiotics to prevent virus fecal transmission. Our study proposed this novel construction strategy of engineering probiotics with both adhesion and neutralization effects, which provided a new therapeutic view for intestinal virus clearance.

Cross-device federated learning (FL) protects user privacy by collaboratively training a model on user devices, therefore eliminating the need for collecting, storing, and manually labeling user data. While important topics such as the FL training algorithm, non-IID-ness, and Differential Privacy have been well studied in the literature, this paper focuses on two challenges of practical importance for improving on-device ASR: the lack of ground-truth transcriptions and the scarcity of compute resource and network bandwidth on edge devices. First, we propose a FL system for on-device ASR domain adaptation with full self-supervision, which uses self-labeling together with data augmentation and filtering techniques. The system can improve a strong Emformer-Transducer based ASR model pretrained on out-of-domain data, using in-domain audio without any ground-truth transcriptions. Second, to reduce the training cost, we propose a self-restricted RNN Transducer (SR-RNN-T) loss, a variant of alignment-restricted RNN-T that uses Viterbi alignments from self-supervision. To further reduce the compute and network cost, we systematically explore adapting only a subset of weights in the Emformer-Transducer. Our best training recipe achieves a \\(12.9\\%\\) relative WER reduction over the strong out-of-domain baseline, which equals \\(70\\%\\) of the reduction achievable with full human supervision and centralized training.

Measuring automatic speech recognition (ASR) system quality is critical for creating user-satisfying voice-driven applications. Word Error Rate (WER) has been traditionally used to evaluate ASR system quality; however, it sometimes correlates poorly with user perception/judgement of transcription quality. This is because WER weighs every word equally and does not consider semantic correctness which has a higher impact on user perception. In this work, we propose evaluating ASR output hypotheses quality with SemDist that can measure semantic correctness by using the distance between the semantic vectors of the reference and hypothesis extracted from a pre-trained language model. Our experimental results of 71K and 36K user annotated ASR output quality show that SemDist achieves higher correlation with user perception than WER. We also show that SemDist has higher correlation with downstream Natural Language Understanding (NLU) tasks than WER.

We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available corpus of speech. The audio has been segmented using voice activity detection and is tagged with SNR, speaker ID and genre descriptions. Additionally, we provide baseline systems and evaluation metrics working under three settings: (1) the zero resource/unsupervised setting (ABX), (2) the semi-supervised setting (PER, CER) and (3) the distant supervision setting (WER). Settings (2) and (3) use limited textual resources (10 minutes to 10 hours) aligned with the speech. Setting (3) uses large amounts of unaligned text. They are evaluated on the standard LibriSpeech dev and test sets for comparison with the supervised state-of-the-art.