Physics, Structures, and Applications of Fluorite-Structured Ferroelectric Tunnel Junctions.

The interest in ferroelectric tunnel junctions (FTJ) has been revitalized by the discovery of ferroelectricity in fluorite-structured oxides such as HfO2 and ZrO2 . In terms of thickness scaling, CMOS compatibility, and 3D integration, these fluorite-structured FTJs provide a number of benefits over conventional perovskite-based FTJs. Here, recent developments involving all FTJ devices with fluorite structures are examined. The transport mechanism of fluorite-structured FTJs is explored and contrasted with perovskite-based FTJs and other 2-terminal resistive switching devices starting with the operation principle and essential parameters of the tunneling electroresistance effect. The applications of FTJs, such as neuromorphic devices, logic-in-memory, and physically unclonable function, are then discussed, along with several structural approaches to fluorite-structure FTJs. Finally, the materials and device integration difficulties related to fluorite-structure FTJ devices are reviewed. The purpose of this review is to outline the theories, physics, fabrication processes, applications, and current difficulties associated with fluorite-structure FTJs while also describing potential future possibilities for optimization.

End-to-End Lip-Reading Open Cloud-Based Speech Architecture.

Deep learning technology has encouraged research on noise-robust automatic speech recognition (ASR). The combination of cloud computing technologies and artificial intelligence has significantly improved the performance of open cloud-based speech recognition application programming interfaces (OCSR APIs). Noise-robust ASRs for application in different environments are being developed. This study proposes noise-robust OCSR APIs based on an end-to-end lip-reading architecture for practical applications in various environments. Several OCSR APIs, including Google, Microsoft, Amazon, and Naver, were evaluated using the Google Voice Command Dataset v2 to obtain the optimum performance. Based on performance, the Microsoft API was integrated with Google's trained word2vec model to enhance the keywords with more complete semantic information. The extracted word vector was integrated with the proposed lip-reading architecture for audio-visual speech recognition. Three forms of convolutional neural networks (3D CNN, 3D dense connection CNN, and multilayer 3D CNN) were used in the proposed lip-reading architecture. Vectors extracted from API and vision were classified after concatenation. The proposed architecture enhanced the OCSR API average accuracy rate by 14.42% using standard ASR evaluation measures along with the signal-to-noise ratio. The proposed model exhibits improved performance in various noise settings, increasing the dependability of OCSR APIs for practical applications.

End-to-End Sentence-Level Multi-View Lipreading Architecture with Spatial Attention Module Integrated Multiple CNNs and Cascaded Local Self-Attention-CTC.

Concomitant with the recent advances in deep learning, automatic speech recognition and visual speech recognition (VSR) have received considerable attention. However, although VSR systems must identify speech from both frontal and profile faces in real-world scenarios, most VSR studies have focused solely on frontal face pictures. To address this issue, we propose an end-to-end sentence-level multi-view VSR architecture for faces captured from four different perspectives (frontal, 30°, 45°, and 60°). The encoder uses multiple convolutional neural networks with a spatial attention module to detect minor changes in the mouth patterns of similarly pronounced words, and the decoder uses cascaded local self-attention connectionist temporal classification to collect the details of local contextual information in the immediate vicinity, which results in a substantial performance boost and speedy convergence. To compare the performance of the proposed model for experiments on the OuluVS2 dataset, the dataset was divided into four different perspectives, and the obtained performance improvement was 3.31% (0°), 4.79% (30°), 5.51% (45°), 6.18% (60°), and 4.95% (mean), respectively, compared with the existing state-of-the-art performance, and the average performance improved by 9.1% compared with the baseline. Thus, the suggested design enhances the performance of multi-view VSR and boosts its usefulness in real-world applications.

Noise-Robust Multimodal Audio-Visual Speech Recognition System for Speech-Based Interaction Applications.

Speech is a commonly used interaction-recognition technique in edutainment-based systems and is a key technology for smooth educational learning and user-system interaction. However, its application to real environments is limited owing to the various noise disruptions in real environments. In this study, an audio and visual information-based multimode interaction system is proposed that enables virtual aquarium systems that use speech to interact to be robust to ambient noise. For audio-based speech recognition, a list of words recognized by a speech API is expressed as word vectors using a pretrained model. Meanwhile, vision-based speech recognition uses a composite end-to-end deep neural network. Subsequently, the vectors derived from the API and vision are classified after concatenation. The signal-to-noise ratio of the proposed system was determined based on data from four types of noise environments. Furthermore, it was tested for accuracy and efficiency against existing single-mode strategies for extracting visual features and audio speech recognition. Its average recognition rate was 91.42% when only speech was used, and improved by 6.7% to 98.12% when audio and visual information were combined. This method can be helpful in various real-world settings where speech recognition is regularly utilized, such as cafés, museums, music halls, and kiosks.

Effect of Ga composition on mobility in a-InGaZnO thin-film transistors.

Oxide semiconductor TFTs have attracted considerable attention in the recent past due to their excellent mobility, high optical transparency in the visible region, and most importantly their fabrication process at low-temperature. However, charge trapping formation in the gate dielectric and the interfaces in such oxide TFTs leads to serious issues such as their operational stability and reliability. Understanding the charge trapping mechanism is therefore of utmost importance to identify the root cause of the aforesaid problems. In this report, we present a detailed study on the charge trapping and dynamic charge transport of a-IGZO TFTs by examining microsecond fast IV (FIV), pulse IV (PIV), and transient IV measurements. The a-IGZO TFTs have designed and fabricated with various Ga compositions (0, 0.14 and 0.22). It was observed that the charge trapping in the a-IGZO TFT is reliant on the sweeping time and the carrier mobility measured using the FIV technique was found to be higher than that obtained from the conventional DC IV measurement. Mobility values ([Formula: see text]) was also measured through the PIV technique and are found to be approximately 10%, 16%, and 21% lower than the intrinsic mobility values. Temperature-dependent study reveals that the intrinsic mobility values (18.45, 16.1 and 12.03 cm2 V-1 s-1) are higher than the pulse mobility values for various Ga compositions (0, 0.14 and 0.22) at higher temperature (175 °C) probably due to the formation of free carriers. Suitable optimization of process parameters of a-IGZO TFTs can therefore enhance the device stability and reliability characteristics leading to their potential utilization in flexible and stretchable electronic devices, sensors & detectors and biomedical devices.

Lipreading Architecture Based on Multiple Convolutional Neural Networks for Sentence-Level Visual Speech Recognition.

In visual speech recognition (VSR), speech is transcribed using only visual information to interpret tongue and teeth movements. Recently, deep learning has shown outstanding performance in VSR, with accuracy exceeding that of lipreaders on benchmark datasets. However, several problems still exist when using VSR systems. A major challenge is the distinction of words with similar pronunciation, called homophones; these lead to word ambiguity. Another technical limitation of traditional VSR systems is that visual information does not provide sufficient data for learning words such as "a", "an", "eight", and "bin" because their lengths are shorter than 0.02 s. This report proposes a novel lipreading architecture that combines three different convolutional neural networks (CNNs; a 3D CNN, a densely connected 3D CNN, and a multi-layer feature fusion 3D CNN), which are followed by a two-layer bi-directional gated recurrent unit. The entire network was trained using connectionist temporal classification. The results of the standard automatic speech recognition evaluation metrics show that the proposed architecture reduced the character and word error rates of the baseline model by 5.681% and 11.282%, respectively, for the unseen-speaker dataset. Our proposed architecture exhibits improved performance even when visual ambiguity arises, thereby increasing VSR reliability for practical applications.

Broad spectral responsivity in highly photoconductive InZnO/MoS2 heterojunction phototransistor with ultrathin transparent metal electrode.

An amorphous InZnO/MoS2 heterojunction-based phototransistor with excellent photoconductive gain and responsivity over the entire visible range has been demonstrated. The photogenerated current of the InZnO phototransistor at long light wavelength (>600 nm) was significantly improved by utilizing narrow bandgap MoS2 as the capping layer (1.3 eV). At lower wavelength, photocarriers are generated due to the optical absorption of both InZnO and MoS2 layers, whereas the latter ensures significant photocarrier generation even at the higher wavelength region of the visible spectrum. The photogenerated carriers subsequently transfer to the underlying InZnO layer of superior carrier mobility that has a high channel conduction of additional electrons from the optically-induced doubly positively charged oxygen vacancies (Vo++) where the gate field is screening, thereby leading to the higher photoconductive gain of the InZnO/MoS2 phototransistors. The dynamic photosensitivity behaviour of the aforesaid phototransistor reveals the presence of persistent photoconductivity (PPC) due to the oxygen vacancy associated with InZnO which can be removed by applying a reset gate pulse from -15 to +5 V. The optical properties of these phototransistors were further enhanced by replacing the opaque Ti/Au electrode by an ultrathin transparent Ti/Au electrode. Utilization of the transparent electrode results in enhanced electron injection from source to channel due to a reduced barrier height under illumination giving rise to a ten-fold improvement in the photocurrent and responsivity of the phototransistors. A position-dependent study of the photocurrent w.r.t beam position also reveals that the enhancement in photocurrent is strongly dependent on the position and is at its maximum when the beam is placed near the source region.

Comprehensive study of high pressure annealing on the ferroelectric properties of Hf0.5Zr0.5O2 thin films.

Thin films of ferroelectric materials are potential candidates to be implemented in the unfolding of a new paradigm in high-density memory devices. As the thickness of these films reaches the sub-10 nm level, the interface properties between the electrode and ferroelectric material undergo significant changes that play a crucial role in governing the ferroelectric behavior. The present state-of-the-art approach presents a detailed investigation of different high pressure annealing (HPA) conditions through simulation studies. The simulation studies were performed using Landau-Khalatnikov equations, with Landau's parameters calculated using the least regression method as described in the Method S1. The extracted coefficients were used to determine various relationships (free energy, ferroelectric potential and negative capacitance) with which to observe the impact of HPA on the negative capacitance (NC) effect on account of the majority ferroelectric phase. To verify the simulation results, pulse transient switching measurements were conducted using Pt/Ti/TiN/Hf0.5Zr0.5O2/TiN-based metal-ferroelectric-metal (MFM) devices to study the coercive field, interfacial capacitance and load resistance behavior. The results suggest that the non-ferroelectric portion (t-phase) coexists with the ferroelectric (o-phase) within the thin layer of the MFM capacitor adjacent to TiN electrode, which undergoes a phase transformation from the t-phase to the o-phase when exposed to different HPA conditions as well as electric field cycling during PS measurements. The simulation and experimental results confirm that the 550 °C at 50 atm N2 environment provides the best possibility of achieving the highest ferroelectric characteristics with the lowest proportion of the non-ferroelectric phase and thus the maximum NC effect as well.

The effect of the bottom electrode on ferroelectric tunnel junctions based on CMOS-compatible HfO2.

Ferroelectric tunnel junctions (FTJs) have attracted research interest as promising candidates for non-destructive readout non-volatile memories. Unlike conventional perovskite FTJs, hafnia FTJs offer many advantages in terms of scalability and CMOS compatibility. However, so far, hafnia FTJs have shown poor endurance and relatively low resistance ratios and these have remained issues for real device applications. In our study, we fabricated HfZrO(HZO)-based FTJs with various electrodes (TiN, Si, SiGe, Ge) and improved the memory performance of HZO-based FTJs by using the asymmetry of the charge screening lengths of the electrodes. For the HZO-based FTJ with a Ge substrate, the effective barrier afforded by this FTJ can be electrically modulated because of the space charge-limited region formed at the ferroelectric/semiconductor interface. The optimized HZO-based FTJ with a Ge bottom electrode presents excellent ferroelectricity with a high remnant polarization of 18 µC cm-2, high tunneling electroresistance value of 30, good retention at 85 °C and high endurance of 107. The results demonstrate the great potential of HfO2-based FTJs in non-destructive readout non-volatile memories.

Quantitative analysis of charge trapping and classification of sub-gap states in MoS2 TFT by pulse I-V method.

The threshold voltage instabilities and huge hysteresis of MoS2 thin film transistors (TFTs) have raised concerns about their practical applicability in next-generation switching devices. These behaviors are associated with charge trapping, which stems from tunneling to the adjacent trap site, interfacial redox reaction and interface and/or bulk trap states. In this report, we present quantitative analysis on the electron charge trapping mechanism of MoS2 TFT by fast pulse I-V method and the space charge limited current (SCLC) measurement. By adopting the fast pulse I-V method, we were able to obtain effective mobility. In addition, the origin of the trap states was identified by disassembling the sub-gap states into interface trap and bulk trap states by simple extraction analysis. These measurement methods and analyses enable not only quantitative extraction of various traps but also an understanding of the charge transport mechanism in MoS2 TFTs. The fast I-V data and SCLC data obtained under various measurement temperatures and ambient show that electron transport to neighboring trap sites by tunneling is the main charge trapping mechanism in thin-MoS2 TFTs. This implies that interfacial traps account for most of the total sub-gap states while the bulk trap contribution is negligible, at approximately 0.40% and 0.26% in air and vacuum ambient, respectively. Thus, control of the interface trap states is crucial to further improve the performance of devices with thin channels.

Effect of dysprosium and lutetium metal buffer layers on the resistive switching characteristics of Cu-Sn alloy-based conductive-bridge random access memory.

The conductive-bridge random access memory (CBRAM) has become one of the most suitable candidates for non-volatile memory in next-generation information and communication technology. The resistive switching (RS) mechanism of CBRAM depends on the formation/annihilation of the conductive filament (CF) between the active metal electrode and the inert electrode. However, excessive ion injection from the active electrode into the solid electrolyte reduces the uniformity and reliability of the RS devices. To solve this problem, we investigated the RS characteristics of a CuSn alloy active electrode with different compositions of Cux-Sn1-x (0.13 < X < 0.55). The RS characteristics were further improved by inserting a dysprosium (Dy) or lutetium (Lu) buffer layer at the interface of Cux-Sn1-x/Al2O3. Electrical analysis of the optimal Cu0.4-Sn0.73/Lu-based CBRAM exhibited stable RS behavior with low operation voltage (SET: 0.7 V and RESET: -0.3 V), a high on state/off state resistive ratio (106), AC cyclic endurance (>104), and stable retention (85 °C/10 years). To achieve these performance parameters, CFs were locally formed inside the electrolyte using a modified CuSn active electrode, and the amount of Cu-ion injection was reduced by inserting the Dy or Lu buffer layer between the CuSn active electrode and the electrolyte. In particular, conductive-atomic force microscopy results at the Dy or Lu/Al2O3 interface directly showed and defined the diameter of the CF.

Enhancement of resistive switching properties in Al2O3 bilayer-based atomic switches: multilevel resistive switching.

Atomic switches are considered to be building blocks for future non-volatile data storage and internet of things. However, obtaining device structures capable of ultrahigh density data storage, high endurance, and long data retention, and more importantly, understanding the switching mechanisms are still a challenge for atomic switches. Here, we achieved improved resistive switching performance in a bilayer structure containing aluminum oxide, with an oxygen-deficient oxide as the top switching layer and stoichiometric oxide as the bottom switching layer, using atomic layer deposition. This bilayer device showed a high on/off ratio (105) with better endurance (∼2000 cycles) and longer data retention (104 s) than single-oxide layers. In addition, depending on the compliance current, the bilayer device could be operated in four different resistance states. Furthermore, the depth profiles of the hourglass-shaped conductive filament of the bilayer device was observed by conductive atomic force microscopy.

Determination of intrinsic mobility of a bilayer oxide thin-film transistor by pulsed I-V method.

Amorphous oxide semiconductor thin-film transistors (TFT) have been considered as outstanding switch devices owing to their high mobility. However, because of their amorphous channel material with a certain level of density of states, a fast transient charging effect in an oxide TFT occurs, leading to an underestimation of the mobility value. In this paper, the effects of the fast charging of high-performance bilayer oxide semiconductor TFTs on mobility are examined in order to determine an accurate mobility extraction method. In addition, an approach based on a pulse I D -V G measurement method is proposed to determine the intrinsic mobility value. Even with the short pulse I D -V G measurement, a certain level of fast transient charge trapping cannot be avoided as long as the charge-trap start time is shorter than the pulse rising time. Using a pulse-amplitude-dependent threshold voltage characterization method, we estimated a correction factor for the apparent mobility, thus allowing us to determine the intrinsic mobility.

Pulse I-V characterization of a nano-crystalline oxide device with sub-gap density of states.

Understanding the charge trapping nature of nano-crystalline oxide semiconductor thin film transistors (TFTs) is one of the most important requirements for their successful application. In our investigation, we employed a fast-pulsed I-V technique for understanding the charge trapping phenomenon and for characterizing the intrinsic device performance of an amorphous/nano-crystalline indium-hafnium-zinc-oxide semiconductor TFT with varying density of states in the bulk. Because of the negligible transient charging effect with a very short pulse, the source-to-drain current obtained with the fast-pulsed I-V measurement was higher than that measured by the direct-current characterization method. This is because the fast-pulsed I-V technique provides a charge-trap free environment, suggesting that it is a representative device characterization methodology of TFTs. In addition, a pulsed source-to-drain current versus time plot was used to quantify the dynamic trapping behavior. We found that the charge trapping phenomenon in amorphous/nano-crystalline indium-hafnium-zinc-oxide TFTs is attributable to the charging/discharging of sub-gap density of states in the bulk and is dictated by multiple trap-to-trap processes.

Effect of hydrogen on dynamic charge transport in amorphous oxide thin film transistors.

Hydrogen in zinc oxide based semiconductors functions as a donor or a defect de-activator depending on its concentration, greatly affecting the device characteristics of oxide thin-film transistors (TFTs). Thus, controlling the hydrogen concentration in oxide semiconductors is very important for achieving high mobility and minimizing device instability. In this study, we investigated the charge transport dynamics of the amorphous semiconductor InGaZnO at various hydrogen concentrations as a function of the deposition temperature of the gate insulator. To examine the nature of dynamic charge trapping, we employed short-pulse current-voltage and transient current-time measurements. Among various examined oxide devices, that with a high hydrogen concentration exhibits the best performance characteristics, such as high saturation mobility (10.9 cm(2) v(-1) s(-1)), low subthreshold slope (0.12 V/dec), and negligible hysteresis, which stem from low defect densities and negligible transient charge trapping. Our finding indicates that hydrogen atoms effectively passivate the defects in subgap states of the bulk semiconductor, minimizing the mobility degradation and threshold voltage instability. This study indicates that hydrogen plays a useful role in TFTs by improving the device performance and stability.

Hot Carrier Trapping Induced Negative Photoconductance in InAs Nanowires toward Novel Nonvolatile Memory.

We report a novel negative photoconductivity (NPC) mechanism in n-type indium arsenide nanowires (NWs). Photoexcitation significantly suppresses the conductivity with a gain up to 10(5). The origin of NPC is attributed to the depletion of conduction channels by light assisted hot electron trapping, supported by gate voltage threshold shift and wavelength-dependent photoconductance measurements. Scanning photocurrent microscopy excludes the possibility that NPC originates from the NW/metal contacts and reveals a competing positive photoconductivity. The conductivity recovery after illumination substantially slows down at low temperature, indicating a thermally activated detrapping mechanism. At 78 K, the spontaneous recovery of the conductance is completely quenched, resulting in a reversible memory device, which can be switched by light and gate voltage pulses. The novel NPC based optoelectronics may find exciting applications in photodetection and nonvolatile memory with low power consumption.

Dislocation scatterings in p-type Si(1-x)Ge(x) under weak electric field.

We present a theoretical model which describes hole mobility degradation by charged dislocations in p-type Si(1-x)Ge(x). The complete analytical expression of the dislocation mobility is calculated from the momentum relaxation time of hole carriers under weak electric field. The obtained dislocation mobility shows a T(3/2)/λ relation and is proportional to the germanium density x. We also suggest a criterion for negating scatterings by dislocations in terms of the controllable parameters such as acceptor dopant density, dislocation density, temperature, and Ge density x, etc.

Graphene and thin-film semiconductor heterojunction transistors integrated on wafer scale for low-power electronics.

Graphene heterostructures in which graphene is combined with semiconductors or other layered 2D materials are of considerable interest, as a new class of electronic devices has been realized. Here we propose a technology platform based on graphene-thin-film-semiconductor-metal (GSM) junctions, which can be applied to large-scale and power-efficient electronics compatible with a variety of substrates. We demonstrate wafer-scale integration of vertical field-effect transistors (VFETs) based on graphene-In-Ga-Zn-O (IGZO)-metal asymmetric junctions on a transparent 150 × 150 mm(2) glass. In this system, a triangular energy barrier between the graphene and metal is designed by selecting a metal with a proper work function. We obtain a maximum current on/off ratio (Ion/Ioff) up to 10(6) with an average of 3010 over 2000 devices under ambient conditions. For low-power logic applications, an inverter that combines complementary n-type (IGZO) and p-type (Ge) devices is demonstrated to operate at a bias of only 0.5 V.

Monolithically Integrated Complementary Ferroelectric FET XNOR Synapse for the Binary Neural Network.

Neuromorphic computing, which mimics the structure and principles of the human brain, has the potential to facilitate the hardware implementation of next-generation artificial intelligence systems and process large amounts of data with very low power consumption. Among them, the XNOR synapse-based Binary Neural Network (BNN) has been attracting attention due to its compact neural network parameter size and low hardware cost. The previous XNOR synapse has drawbacks, such as a trade-off between cell density and accuracy. In this work, we show nonvolatile XNOR synapses with high density and accuracy using a monolithically stacked complementary ferroelectric field-effect transistor (C-FeFET) composed of a p-type Si MFMIS-FeFET at the bottom and a 3D stackable n-type Al:IZTO MFS-FeTFT, achieving 60F2 per cell (2C-FeFET). For adjusting the threshold voltage and improving the switching speed (100 ns) of n-type ferroelectric TFT, we employed a dual-gate configuration and a unique operation scheme, making it comparable to those of Si-based FeFETs. We performed array-level simulation with a 512 × 512 subarray size and a 3-bit flash ADC, demonstrating that the image recognition accuracies using the MNIST and CIFAR-10 data sets were increased by 3.17 and 14.07%, respectively, in comparison to other nonvolatile XNOR synapses. In addition, we performed system-level analysis on a 512 × 512 XNOR C-FeFET, exhibiting an outstanding throughput of 717.37 GOPS and an energy efficiency of 196.7 TOPS/W. We expect that our approach would contribute to the high-density memory systems, logic-in-memory technology, and hardware implementation of neural networks.

Gated three-terminal device architecture to eliminate persistent photoconductivity in oxide semiconductor photosensor arrays.

The composition of amorphous oxide semiconductors, which are well known for their optical transparency, can be tailored to enhance their absorption and induce photoconductivity for irradiation with green, and shorter wavelength light. In principle, amorphous oxide semiconductor-based thin-film photoconductors could hence be applied as photosensors. However, their photoconductivity persists for hours after illumination has been removed, which severely degrades the response time and the frame rate of oxide-based sensor arrays. We have solved the problem of persistent photoconductivity (PPC) by developing a gated amorphous oxide semiconductor photo thin-film transistor (photo-TFT) that can provide direct control over the position of the Fermi level in the active layer. Applying a short-duration (10 ns) voltage pulse to these devices induces electron accumulation and accelerates their recombination with ionized oxygen vacancy sites, which are thought to cause PPC. We have integrated these photo-TFTs in a transparent active-matrix photosensor array that can be operated at high frame rates and that has potential applications in contact-free interactive displays.