Channel Layer Research Articles

A Filtering Method for LiDAR Point Cloud Based on Multi-Scale CNN with Attention Mechanism

Point cloud filtering is an important prerequisite for three-dimensional surface modeling with high precision based on LiDAR data. To cope with the issues of low filtering accuracy or excessive model complexity in traditional filtering algorithms, this paper proposes a filtering method for LiDAR point cloud based on a multi-scale convolutional neural network incorporated with the attention mechanism. Firstly, a regular image patch centering on each point is constructed based on the elevation information of point clouds. As thus, the point cloud filtering problem is transformed into the image classification problem. Then, considering the ability of multi-scale convolution to extract features at different scales and the potential of the attention mechanism to capture key information in images, a multi-scale convolutional neural network framework is constructed, and the attention mechanism is incorporated to coordinate multi-scale convolution kernel with channel and spatial attention modules. After this, the feature maps of the LiDAR point clouds can be acquired at different scales. For these feature maps, the weights of each channel layer and different spatial regions can be further tuned adaptively, which makes the network training more targeted, thereby improving the model performance for image classification and eventually separating of ground points and non-ground points preferably. Finally, the proposed method is compared with the cloth simulation filtering method (CSF), deep neural network method (DNN), k-nearest neighbor method (KNN), deep convolutional neural network method (DCNN) and scale-irrelevant and terrain-adaptive method (SITA) for the standard ISPRS dataset of point cloud filtering and the filter dataset of Qinghai. The experimental results show that the proposed method can obtain lower classification errors, which proves the superiority of this method in point cloud filtering.

Impact of layer thickness on the operating characteristics of In2O3/ZnO heterojunction thin-film transistors

Combining low-dimensional layers of dissimilar metal oxide materials to form a heterojunction structure offers a potent strategy to improve the performance and stability of thin-film transistors (TFTs). Here, we study the impact of channel layer thicknesses on the operating characteristics of In2O3/ZnO heterojunction TFTs prepared via sputtering. The conduction band offset present at the In2O3/ZnO heterointerface affects the device's operating characteristics, as is the thickness of the individual oxide layers. The latter is investigated using a variety of experimental and computational modeling techniques. An average field-effect mobility (μFE) of >50 cm2 V−1 s−1, accompanied by a low threshold voltage and a high on/off ratio (∼108), is achieved using an optimal channel configuration. The high μFE in these TFTs is found to correlate with the presence of a quasi-two-dimensional electron gas at the In2O3/ZnO interface. This work provides important insight into the operating principles of heterojunction metal oxide TFTs, which can aid further developments.

Smart Network Maintenance in an Edge Cloud Computing Environment: An Adaptive Model Compression Algorithm Based on Model Pruning and Model Clustering

With the rapid development of communication networks, there are more stringent requirements for maintenance management. To carry out smart network maintenance automatically and effectively, this paper uses wearable devices, robots, and Unmanned Aerial Vehicles (UAVs) to collect on-site video data and detect defects in real-time. However, the deep learning models deployed in these devices should have fewer model parameters and less storage space. Therefore, we propose a model compression algorithm based on model pruning and model clustering in smart network maintenance. First, model pruning combines channel pruning and layer pruning and uses deep reinforcement learning to determine the pruning ratio of each layer automatically. This method can effectively compress the width and depth of the model while maintaining the accuracy of the model. Second, although the pruning operation greatly reduces the redundancy of the number of weight parameters, the number of bits of floating-point weights is still redundant. We propose an adaptive model clustering method to cluster the remaining nonzero parameter weights and compress the model. It combines advanced balanced iterative reducing and clustering using hierarchies (BIRCH) clustering and K-meansII clustering and takes the result k value of BIRCH clustering as the input of K-meansII. It can avoid limitations of prior knowledge and reduce clustering time. Simulation results show that the target detection model can reduce parameter redundancy, save storage space, and simplify calculation using the proposed algorithm. In addition, it can be better applied in the smart network maintenance environment.

Accelerating Bayesian Neural Networks via Algorithmic and Hardware Optimizations

Bayesian neural networks (BayesNNs) have demonstrated their advantages in various safety-critical applications, such as autonomous driving or healthcare, due to their ability to capture and represent model uncertainty. However, standard BayesNNs require to be repeatedly run because of Monte Carlo sampling to quantify their uncertainty, which puts a burden on their real-world hardware performance. To address this performance issue, this article systematically exploits the extensive structured sparsity and redundant computation in BayesNNs. Different from the unstructured or structured sparsity in standard convolutional NNs, the structured sparsity of BayesNNs is introduced by Monte Carlo Dropout and its associated sampling required during uncertainty estimation and prediction, which can be exploited through both algorithmic and hardware optimizations. We first classify the observed sparsity patterns into three categories: channel sparsity, layer sparsity and sample sparsity. On the algorithmic side, a framework is proposed to automatically explore these three sparsity categories without sacrificing algorithmic performance. We demonstrated that structured sparsity can be exploited to accelerate CPU designs by up to 49 times, and GPU designs by up to 40 times. On the hardware side, a novel hardware architecture is proposed to accelerate BayesNNs, which achieves a high hardware performance using the runtime adaptable hardware engines and the intelligent skipping support. Upon implementing the proposed hardware design on an FPGA, our experiments demonstrated that the algorithm-optimized BayesNNs can achieve up to 56 times speedup when compared with unoptimized Bayesian nets. Comparing with the optimized GPU implementation, our FPGA design achieved up to 7.6 times speedup and up to 39.3 times higher energy efficiency.

Transmission properties of grating-gate GaN-based HEMTs under different incident angles in the mid-infrared region.

The absorption tunability of grating-gate GaN-based HEMTs in the mid-infrared region has been confirmed in wide frequency regions. However, the application potential of grating-gate GaN-based HEMTs is limited due to a lack of study on transmission properties under different incident angles. Therefore, this paper studied the transmission characteristics of grating-gate GaN-based HEMTs under different incident angles in the mid-infrared region. By using the optical transfer matrix approach to model the dispersion characteristics in the structure, we found that the stronger plasmon polaritons and phonon polaritons occur in conductive channel and GaN layer. The variation of different incident wave vectors with incident angle affects the plasmon polaritons and phonon polaritons excitation intensities, resulting in the angular tunability transmission properties of grating-gate GaN-based HEMTs. After simulating the electric field distribution in COMSOL, the different transmission properties of grating-gate GaN-based HEMTs occur under different incident angles. Simulated results reveal the excellent angle-selectivity in grating-gate GaN-based HEMTs. The research into these characteristics shows that the structure has a lot of promise for designing mid-infrared angle selection filters, sensors, and other subwavelength devices in the future.

Modulation Doping of Single-Layer Semiconductors for Improved Contact at Metal Interfaces.

Single layers of two-dimensional (2D) materials hold the promise for further miniaturization of semiconductor electronic devices. However, the metal-semiconductor contact resistance limits device performance. To mitigate this problem, we propose modulation doping, specifically a doping layer placed on the opposite side of a metal-semiconductor interface. Using first-principles calculations to obtain the band alignment, we show that the Schottky barrier height and, consequently, the contact resistance at the metal-semiconductor interface can be reduced by modulation doping. We demonstrate the feasibility of this approach for a single-layer tungsten diselenide (WSe2) channel and 2D MXene modulation doping layers, interfaced with several different metal contacts. Our results indicate that the Fermi level of the metal can be shifted across the entire band gap. This approach can be straight-forwardly generalized for other 2D semiconductors and a wide variety of doping layers.

Amorphous InGaZnO Thin-Film Transistors with Double-Stacked Channel Layers for Ultraviolet Light Detection.

Amorphous InGaZnO thin film transistors (a-IGZO TFTs) with double-stacked channel layers (DSCL) were quite fit for ultraviolet (UV) light detection, where the best DSCL was prepared by the depositions of oxygen-rich (OR) IGZO followed by the oxygen-deficient (OD) IGZO films. We investigated the influences of oxygen partial pressure (PO) for DSCL-TFTs on their sensing abilities by experiments as well as Technology Computer Aided Design (TCAD) simulations. With the increase in PO values for the DSCL depositions, the sensing parameters, including photogenerated current (Iphoto), sensitivity (S), responsivity (R), and detectivity (D*) of the corresponding TFTs, apparently degraded. Compared with PO variations for the OR-IGZO films, those for the OD-IGZO depositions more strongly influenced the sensing performances of the DSCL-TFT UV light detectors. The TCAD simulations showed that the variations of the electron concentrations (or oxygen vacancy (VO) density) with PO values under UV light illuminations might account for these experimental results. Finally, some design guidelines for DSCL-TFT UV light detectors were proposed, which might benefit the potential applications of these novel semiconductor devices.

High-performance a-ITZO TFTs with high bias stability enabled by self-aligned passivation using a-GaOx

Maintaining gate bias stability under negative bias stress (NBS) and positive bias stress (PBS) is a long-standing issue in amorphous oxide semiconductor thin-film transistors (TFTs). The passivation of the channel layer is crucial for improving device stability. We show that amorphous gallium oxide, which possesses appropriate energy levels (lower electron affinity and higher ionization potential) for indium–tin–zinc oxide (ITZO) TFTs, can be etched selectively by tetramethyl ammonium hydroxide-containing developers that enable self-alignment passivation, such as easy contact hole formation during the drain and source lithography processes. The self-aligned passivation process led to a-ITZO TFTs with high mobility (>50 cm2 V−1 s−1) and low subthreshold swing (<90 mV/dec). The threshold voltage shifts under NBS and PBS using a bias gate voltage of ±20 V for 1 h were −0.09 and 0.15 V, respectively. This passivation can obviate the need for the conventional CVD-derived passivation process by utilizing the DC sputtering of gallium oxide, which may reduce hydrogen issues.

Artificial neural network-based wall-modeled large-eddy simulations of turbulent channel and separated boundary layer flows

Wall-models in a large-eddy simulation (LES) are essential to alleviate the large near-wall resolution requirements for high-Reynolds-number turbulent flow simulations. Among the existing wall-models for a LES, an equilibrium wall-stress model has the highest computational efficiency. Because this model has limitations, such as a lack of non-equilibrium effects and the assumption of a particular law of the wall in the mean velocity, we propose artificial neural network-based wall-stress models (AWMs). The input variables for the AWMs are extracted from the decomposition of the skin-friction coefficient proposed by Fukagata et al. [1], and the AWMs are shown to be able to predict the wall-shear stress in complex flows accurately. The performance of the AWMs is tested for two types of flows, a fully developed turbulent channel flow and a separated turbulent boundary layer flow. A direct comparison of the turbulence statistics with those obtained by previous wall-models (i.e., a log-law-based wall-stress model and a non-equilibrium wall-stress model) shows that better predictions are achieved using the AWMs for both flows, even with untrained Reynolds numbers. When using a coarse grid along the wall-normal direction in wall-modeled LESs (WMLESs) with the AWMs, an upward shift of the mean velocity profile (positive log-layer mismatch, LLM) compared to direct numerical simulation data is found, consistent with previous studies. However, this LLM problem can be overcome by imposing a filtered wall-normal velocity at the wall that is dynamically determined based on the continuity equation and the Taylor series expansion within wall-adjacent cells.

Fatigue Effect of Stretchable a-InGaZnO TFT on PI/PDMS Substrate under Repetitive Uni/Biaxial Elongation Stress

We examine the fatigue effects of repeated elongation stress for a fabricated a-IGZO thin-film transistor (TFT) on an island polyimide (PI)/poly(dimethylsiloxane) (PDMS) substrate. The repeated elongation stress evaluation was performed in both a uniaxial direction with 10, 30, and 50% strain and biaxial direction with 10, 20, and 30% strain for 50 000 cycles. The a-IGZO TFTs on the island PI exhibit the following parameters: Vth: −0.91 ± 0.30 V, μsat: 26.3 ± 0.61 cm2/(V s), and SS: 0.40 ± 0.02 V/decade. As a result of the repeated elongation stress in the maximum evaluated elongation (uniaxis: 50% and biaxis: 30%), the subthreshold swing (SS) increased by about 0.20 V/decade in all directions, and the saturation mobility decreased only in the x-axis direction (Δμsat: 26.30 → 21.43 cm2/(V s)). Through XPS analysis and TCAD simulation, it was confirmed that SS deteriorated due to an increase of the oxygen-vacancy-related defects (NGA and NGD) in the IGZO channel layer after repeated elongation stress. The transmission line method (TLM) confirmed that contact resistance increased when stretched only in the x-axis direction. In addition, the positive bias stress (PBS) test showed abnormal behavior including an additional SS-degradation-induced Vth negative shift after repeated elongation stress in all directions. Consequently, the degradation mechanism of both the electrical properties and bias stability for the stretchable a-IGZO TFT was determined after elongation stresses.

Improved electrical performance of lateral β-Ga2O3 MOSFETs utilizing slanted fin channel structure

In this Letter, lateral slanted-fin-channel β-Ga2O3 metal-oxide-semiconductor field effect transistors (MOSFETs) are demonstrated. A 600-nm thick n-type doped channel layer is adopted to improve output characteristics. The tri-gate structure enhances gate control in the proposed β-Ga2O3 MOSFETs, showing an on/off ratio as high as 109. In particular, the slanted-fin-channel structure, mainly located in the gate region, reduces the peak electric field in the Ga2O3 channel due to the gradual regulation of a threshold voltage. The slanted-fin-channel β-Ga2O3 MOSFETs show a breakdown voltage (Vbr) of 2400 V and a power figure-of-merit of 193 MW/cm2, which are almost 2 and 5.5 times larger, respectively, than those of conventional straight-fin-channel devices. These results imply that the slanted-fin channel structure provides a viable way of fabricating high-performance β-Ga2O3 MOSFET power devices.

Developing Subthreshold-Swing Limit of PEALD In–Sn–Ga–O Transistor via Atomic-Scaled Sn Control

The In–Sn–Ga–O (ITGO) thin-film transistor (TFT) is promising in that it possesses enhanced electrical characteristics and stability because the tin (Sn) has large spherical s orbitals and a high binding energy with oxygen (O). Recently, there have been several reports of ITGO material fabricated via sputtering. Therefore, studies that control Sn composition to achieve unique characteristics and obtain conformal films have been limited. For these reasons, we evaluated plasma-enhanced atomic layer deposition (PEALD)-derived ITGO materials with varied Sn composition. The electrical characteristics are enhanced with the Sn subcycle ratio from 1 to 5 in the ITGO films (carrier concentration: 6.4 × 1020 to 2.2 × 1021 cm–3, resistivity: 8.6 × 10–4 to 6.1 × 10–4 Ω cm). These results are due to carrier generation by the Sn, a decrease in oxygen-related defects (22.9% to 19.5%), and suppressed indium-oxide (In2O3) crystallinity. To demonstrate their applicability to practical devices, the ITGO films were applied to TFTs as active channel layers. The ITGO TFTs with increasing Sn content show a decrease then increase of subthreshold swing (S.S.) characteristics, while the field-effect mobility (μFE) slightly increases. In addition, the ITGO TFTs show improved negative bias illumination stress (NBIS) stability with increasing Sn content, while the turnaround point is observed in the Sn 5 subcycle. As a result, the ITGO TFT with the Sn 3 subcycle shows an extremely low S.S. of 64.8 ± 1.9 mV/decade and an NBIS result of −0.5 V. The outstanding performance of the ITGO TFT with Sn 3 subcycle is attributed to its low number of oxygen-related defects, amorphous structure, and low valence band maximum (VBM) level. Therefore, the ITGO TFTs with optimized Sn content have the advantages of low power consumption and resistance to an environment with illumination.

High performance, amorphous InGaZnO thin-film transistors with ferroelectric ZrO2 gate insulator by one step annealing

Ferroelectric thin film transistors (FE-TFTs) are of increasing interest for next generation memory applications. Post fabrication annealing plays a crucial role to induce ferroelectricity in the gate insulator used for FE-TFT. The one step post fabrication annealing can induce ferroelectric phase in the gate insulator and reduce the defects in the semiconductor channel layer. We report a new technique to induce the ferroelectricity in spray coated ZrO2 with sputtered amorphous InGaZnO (a-IGZO) capping layer by employing one-step annealing. We achieved a large memory window of 1.5 V, high ION/IOFF ratio of 1 × 107, and steep subthreshold swing (SS) of 0.12 V/decade by one step annealing at 550 °C. The low thermal expansion coefficient (≈ 4.31 × 10−6/K) of a-IGZO helps to induce polar orthorhombic phase by providing suitable mechanical stress to the underlying ZrO2 layer. This work provides a new approach to induce ferroelectricity in ZrO2 for high performance ferroelectric thin film transistors.

Numerical analyses on oxygen transport resistances in polymer electrolyte membrane fuel cells using a novel agglomerate model

Characterizing oxygen transport resistances in different components of a polymer electrolyte membrane fuel cell (PEMFC) is essential to achieve better cell performance at high current under low Pt loading. In this work, a macroscopic three-dimensional model, together with a novel agglomerate model was proposed to analyze impacts of operating conditions on these resistances via limiting current strategy. By introducing a focusing factor obtained with lattice Boltzmann method at mesoscopic level, the structure-dependent local transport resistance in ionomer thin-film of the electrode was comprehensively captured and validated by existing experimental studies. Contributions of the cell components to the total transport resistance were dissected. Results show that the present agglomerate model could well reproduce the local transport behaviors of oxygen in catalyst layer by fully considering the detailed nanoscale diffusion and adsorption processes. A small mass fraction of oxygen was favored to minimize the relative deviation of the local transport resistance from its intrinsic one due to the water production and heat generation, which can reach 7% for the mass fraction of oxygen of 1%. Contribution of the in-plane diffusion of oxygen in the inactive electrode is around 1%. The total transport resistance increased with the absolute pressure, mainly due to the dominated molecular diffusion mechanism in gas channel and gas diffusion layer. Gas convection accounted for 26% of the oxygen transport resistance originated from gas channel. The transport resistance of catalyst layer increased significantly with the reduction of Pt loading, and decreased with relative humidity and operating temperature, particularly at high Pt loading.

Impact of the Channel Thickness on Electron Confinement in MOCVD‐Grown High Breakdown Buffer‐Free AlGaN/GaN Heterostructures

The 2D electron gas (2DEG) confinement on high electron mobility transistor (HEMT) heterostructures with a thin undoped GaN channel layer on the top of a grain‐boundary‐free AlN nucleation layer is studied. This is the first time demonstration of a buffer‐free epi‐structure grown with metal–organic chemical vapor deposition with thin GaN channel thicknesses, ranging from 250 to 150 nm, without any degradation of the structural quality and 2DEG properties. The HEMTs with a gate length of 70 nm exhibit good DC characteristics with peak transconductances of 500 mS mm−1and maximum saturated drain currents above 1 A mm−1. A thinner GaN channel layer improves 2DEG confinement because of the enhanced effectiveness of the AlN nucleation layer acting as a back‐barrier. An excellent drain‐induced barrier lowering of only 20 mV V−1at aVDSof 25 V and an outstanding critical electric field of 0.95 MV cm−1are demonstrated. Good large‐signal performance at 28 GHz with output power levels of 2.0 and 3.2 W mm−1and associated power‐added efficiencies of 56% and 40% are obtained at aVDSof 15 and 25 V, respectively. These results demonstrate the potential of sub‐100 nm gate length HEMTs on a buffer‐free GaN‐on‐SiC heterostructure.

Overall enhancement of electrical properties of indium oxide thin-film transistors by surface doping

Doping is a powerful tool for modulating the charge carrier density in metal oxide thin-film transistors (TFTs). However, one crucial limitation is the disruption of microstructure after doping of crystalline semiconductors, which dramatically decreases the charge carrier mobility. Here, we report on a surface doping method that utilizes organic acceptor molecules to efficiently modulate the carrier concentration without disrupting the lattice of thin indium oxide channel layers. For the first time, we achieve electron transfer from indium oxide channel to the organic layer, instead conventional strategy of organic to metal oxide, to control the intrinsic carrier concentration and enhance the electrical performance of the indium oxide TFTs. Moreover, we use a complementary tool based on spectroscopic techniques to prove the fundamental charge transfer processes between metal oxide semiconductors and organic molecular dopants. The results indicate that surface doping is a simple and effective way to improve the performance of oxide TFTs and suitable for potential applications in large-area electronics. • Indium oxide TFTs were fabricated by solution process method. • Surface doping modulates the carrier concentration of the channel layer. • The electrical performances of the devices were enhanced by surface doping.

Cross-Technology Communication for Heterogeneous Wireless Devices Through Symbol-Level Energy Modulation

The coexistence of heterogeneous devices in wireless networks brings a new topic on cross-technology communication (CTC) to improve the coexistence efficiency and boost collaboration among these devices. Current advances on CTC mainly fall into two categories, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">physical-layer CTC</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">packet-level energy modulation</i> (PLEM). The <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">physical-layer CTC</i> achieves a high CTC data rate, but with channel incompatible to commercial devices, making it hard to be deployed in current wireless networks. PLEM is channel and physical layer compatible, but with two main drawbacks of the low CTC data rate and MAC incompatibility, which will induce severe interference to the other devices’ normal data transmissions. In this paper, we propose symbol-level energy modulation (SLEM), the first CTC method that is fully compatible with current devices in both channel and the physical/MAC layer processes, having the ability to be deployed in commercial wireless networks smoothly. SLEM inserts extra bits to WiFi data bits to generate the transmitting bits, so as to adjust the energy levels of WiFi symbols to deliver CTC information. We make theoretical analysis to figure out the performance of both CTC and WiFi transmissions. We also conduct experiments to demonstrate the feasibility of SLEM and its performance under different network situations.

Tuning Ambipolarity of the Conjugated Polymer Channel Layers of Floating‐Gate Free Transistors: From Volatile Memories to Artificial Synapses (Adv. Sci. 31/2022)

Tuning Ambipolarity of the Conjugated Polymer Channel Layers of Floating‐Gate Free Transistors: From Volatile Memories to Artificial Synapses (Adv. Sci. 31/2022)

Three-dimensional CFD analysis of PEMFC with different membrane thicknesses

PEM fuel cell (PEMFC) is a potential candidate for future source of power used in different applications such as transportations, stationary and portable power. PEMFC consists of different parts including membrane, bipolar plate, flow channel, gas diffusion and catalyst layers. Membrane is one of the most important components of a PEMFC and its physical and geometrical features significantly affect PEMFC efficiency. In this paper, a three-dimensional, single-phase computational model has been improved to scrutinize membrane thickness effect on the PEM fuel cell performance using the ANSYS PEM Fuel Cell Module. Membrane thicknesses are in the range of a range of 0.0127 to 0.189 mm. The results reveal that a decrease in membrane thickness augments the current density at 0.4 and 0.6 V. The peak current density of 3.12 A/cm 2 is achieved with 0.027 mm membrane thickness compared the model current density of 1.26 A/cm 2 obtained by the model with 0.128 mm membrane thickness at 0.4 V. Oxygen consumption and water production also enhance with reducing membrane thickness at 0.4 and 0.6 V. However change thickness of the membrane has a negligible impact on impact pressure drop in the flow channel. It is found that optimization of membrane thickness are necessary for attaining high efficiency.

Effect of oxygen partial pressure on the performance of homojunction amorphous In-Ga-Zn-O thin-film transistors

In this study, the homojunction thin-film transistors (TFTs) with amorphous indium gallium zinc oxide (a-IGZO) as active channel layers and source/drain electrodes were fabricated by RF magnetron sputtering. The effect of oxygen partial pressure on the phase, microstructure, optical and electrical properties of IGZO thin films was investigated. The results showed that amorphous IGZO thin films always exhibit a high transmittance above 90% and wide band gaps of around 3.9 eV. The resistivity increases as the IGZO thin films are deposited at a higher oxygen partial pressure due to the depletion of oxygen vacancies. In addition, the electrical behaviors in homojunction IGZO TFTs were analyzed. When the active channel layers were deposited with an oxygen partial pressure of 1.96%, the homojunction IGZO TFTs exhibited optimal transfer and output characteristics with a field-effect mobility of 13.68 cm2 V−1 s−1. Its sub-threshold swing, threshold voltage and on/off ratio are 0.6 V/decade, 0.61 V and 107, respectively.