Innovations of metallic contacts on semiconducting 2D transition metal dichalcogenides toward advanced 3D-structured field-effect transistors.

2D semiconductors, represented by transition metal dichalcogenides (TMDs), have the potential to be alternative channel materials for advanced 3D field-effect transistors, such as gate-all-around field-effect-transistors (GAAFETs) and complementary field-effect-transistors (C-FETs), due to their inherent atomic thinness, moderate mobility, and short scaling lengths. However, 2D semiconductors encounter several technological challenges, especially the high contact resistance issue between 2D semiconductors and metals. This review provides a comprehensive overview of the high contact resistance issue in 2D semiconductors, including its physical background and the efforts to address it, with respect to their applicability to GAAFET structures.

Proformer: a hybrid macaron transformer model predicts expression values from promoter sequences.

The breakthrough high-throughput measurement of the cis-regulatory activity of millions of randomly generated promoters provides an unprecedented opportunity to systematically decode the cis-regulatory logic that determines the expression values. We developed an end-to-end transformer encoder architecture named Proformer to predict the expression values from DNA sequences. Proformer used a Macaron-like Transformer encoder architecture, where two half-step feed forward (FFN) layers were placed at the beginning and the end of each encoder block, and a separable 1D convolution layer was inserted after the first FFN layer and in front of the multi-head attention layer. The sliding k-mers from one-hot encoded sequences were mapped onto a continuous embedding, combined with the learned positional embedding and strand embedding (forward strand vs. reverse complemented strand) as the sequence input. Moreover, Proformer introduced multiple expression heads with mask filling to prevent the transformer models from collapsing when training on relatively small amount of data. We empirically determined that this design had significantly better performance than the conventional design such as using the global pooling layer as the output layer for the regression task. These analyses support the notion that Proformer provides a novel method of learning and enhances our understanding of how cis-regulatory sequences determine the expression values.

Evaluation and optimization of sequence-based gene regulatory deep learning models.

Neural networks have emerged as immensely powerful tools in predicting functional genomic regions, notably evidenced by recent successes in deciphering gene regulatory logic. However, a systematic evaluation of how model architectures and training strategies impact genomics model performance is lacking. To address this gap, we held a DREAM Challenge where competitors trained models on a dataset of millions of random promoter DNA sequences and corresponding expression levels, experimentally determined in yeast, to best capture the relationship between regulatory DNA and gene expression. For a robust evaluation of the models, we designed a comprehensive suite of benchmarks encompassing various sequence types. While some benchmarks produced similar results across the top-performing models, others differed substantially. All top-performing models used neural networks, but diverged in architectures and novel training strategies, tailored to genomics sequence data. To dissect how architectural and training choices impact performance, we developed the Prix Fixe framework to divide any given model into logically equivalent building blocks. We tested all possible combinations for the top three models and observed performance improvements for each. The DREAM Challenge models not only achieved state-of-the-art results on our comprehensive yeast dataset but also consistently surpassed existing benchmarks on Drosophila and human genomic datasets. Overall, we demonstrate that high-quality gold-standard genomics datasets can drive significant progress in model development.

Broadband Van-der-Waals Photodetector Driven by Ferroelectric Polarization.

The potential for various future industrial applications has made broadband photodetectors beyond visible light an area of great interest. Although most 2D van-der-Waals (vdW) semiconductors have a relatively large energy bandgap (>1.2 eV), which limits their use in short-wave infrared detection, they have recently been considered as a replacement for ternary alloys in high-performance photodetectors due to their strong light-matter interaction. In this study, a ferroelectric gating ReS2 /WSe2 vdW heterojunction-channel photodetector is presented that successfully achieves broadband light detection (>1300 nm, expandable up to 2700 nm). The staggered type-II bandgap alignment creates an interlayer gap of 0.46 eV between the valence band maximum (VBMAX ) of WSe2 and the conduction band minimum (CBMIN ) of ReS2 . Especially, the control of poly(vinylidene fluoride-trifluoroethylene) (P(VDF-TrFE)) ferroelectric dipole polarity for a specific wavelength allows a high photoresponsivity of up to 6.9 × 103 A W-1 and a low dark current below 0.26 nA under the laser illumination with a wavelength of 405 nm in P-up mode. The achieved high photoresponsivity, low dark current, and full-range near infrared (NIR) detection capability open the door for next-generation photodetectors beyond traditional ternary alloy photodetectors.

A Van Der Waals Reconfigurable Multi-Valued Logic Device and Circuit Based on Tunable Negative-Differential-Resistance Phenomenon.

Multi-valued logic (MVL) technology that utilizes more than two logic states has recently been reconsidered because of the demand for greater power saving in current binary logic systems. Extensive efforts have been invested in developing MVL devices with multiple threshold voltages by adopting negative differential transconductance and resistance. In this study, a reconfigurable, multiple negative-differential-resistance (m-NDR) device with an electric-field-induced tunability of multiple threshold voltages is reported, which comprises a BP/ReS2 heterojunction and a ReS2 /h-BN/metal capacitor. Tunability for the m-NDR phenomenon is achieved via the resistance modulation of the ReS2 layer by electrical pulses applied to the capacitor region. Reconfigurability is verified in terms of the function of an MVL circuit composed of a reconfigurable m-NDR device and a load transistor, wherein staggered-type and broken-type double peak-NDR device operations are adopted for ternary inverter and latch circuits, respectively.

Two-Dimensional MXene Synapse for Brain-Inspired Neuromorphic Computing.

MXenes, an emerging class of two-dimensional (2D) transition metal carbides and nitrides, have attracted wide attention because of their fascinating properties required in functional electronics. Here, an atomic-switch-type artificial synapse fabricated on Ti3 C2 Tx MXene nanosheets with lots of surface functional groups, which successfully mimics the dynamics of biological synapses, is reported. Through in-depth analysis by X-ray photoelectron spectroscopy, transmission electron microscopy, and energy dispersive X-ray spectroscopy, it is found that the synaptic dynamics originated from the gradual formation and annihilation of the conductive metallic filaments on the MXene surface with distributed functional groups. Subsequently, via training and inference tasks using a convolutional neural network for the Canadian-Institute-For-Advanced-Research-10 dataset, the applicability of the artificial MXene synapse to hardware neural networks is demonstrated.

Evaluation of lake modification alternatives for Lake Sihwa, Korea.

After the construction of an embankment at the Bay of Sihwa in Korea, a lake of 56.5 km2 surface area and 330 million m3 volume was created. Because of rapid socioenvironmental changes and the lowering of water quality in Lake Sihwa, various external measures have been proposed and some of them are being implemented. In this paper, we examine two alternatives for in-lake modification: one alternative is zoning of the lake by constructing two submerged dams and the other is channeling of the lake through reclamation. Water quality modeling was conducted for both alternatives to assess their effects. Results of the modeling revealed that the reduction of the lake size through two different approaches, when accompanied with other external measures, would improve the water quality, but to different degrees. The zonation is expected to improve the freshwater quality up to the level supplying 45 million m3 of water per year for agricultural use. The quality of channeled water would be inappropriate for agricultural use, but suitable for outdoor activities such as recreation or fishing regardless of reclamation plans considered.