Reconfigurable Artificial Synapses with Excitatory and Inhibitory Response Enabled by an Ambipolar Oxide Thin-Film Transistor.

A gate-tunable synaptic device controlling dynamically reconfigurable excitatory and inhibitory synaptic responses, which can emulate the fundamental synaptic responses for developing diverse functionalities of the biological nervous system, was developed using ambipolar oxide semiconductor thin-film transistors (TFTs). Since the balanced ambipolarity is significant, a boron-incorporated SnO (SnO:B) oxide semiconductor channel was newly developed to improve the ambipolar charge transports by reducing the subgap defect density, which was reduced to less than 1017 cm-3. The ambipolar SnO:B-TFT could be fabricated with a good reproductivity at the maximum process temperature of 250 °C and exhibited good TFT performances, such as a nearly zero switching voltage, the saturation mobility of ∼1.3 cm2 V-1 s-1, s-value of ∼1.1 V decade-1, and an on/off-current ratio of ∼8 × 103 for the p-channel mode, while ∼0.14 cm2 V-1 s-1, ∼2.2 V decade-1and ∼1 × 103 for n-channel modes, respectively. The ambipolar device imitated potentiation/depression behaviors in both excitatory and inhibitory synaptic responses by using the p- and n-channel transports by tuning a gate bias. The low-power consumptions of <20 and <2 nJ per pulse for the excitatory and inhibitory operations, respectively, were also achieved. The presented device operated under an ambient atmosphere and confirmed a good operation reliability over 5000 pulses and a long-term air environmental stability. The study presents the high potential of an ambipolar oxide-TFT-based synaptic device with a good manufacturability to develop emerging neuromorphic perception and computing hardware for next-generation artificial intelligence systems.

Vacuum-Free Liquid-Metal-Printed 2D Indium-Tin Oxide Thin-Film Transistor for Oxide Inverters.

A cost-effective, vacuum-free, liquid-metal-printed two-dimensional (2D) (∼1.9 nm-thick) tin-doped indium oxide (ITO) thin-film transistor (TFT) was developed at the maximum process temperature of 200 °C. A large-sized 2D-ITO channel layer with an electron density of ∼1.2 × 1019 cm-3 was prepared in an ambient atmosphere. The 2D-ITO-TFT operated in full depletion with a threshold voltage of -2.1 V and demonstrated good TFT device characteristics such as a high saturation mobility of ∼27 cm2 V-1 s-1, a small subthreshold slope of <382 mV decade-1, and a large on/off-current ratio of >109. The TFT device simulation analysis found that the 2D-ITO-TFT performances were controlled by the shallow acceptor-like in-gap defects spreading in the midgap region of over 1.0 eV below the conduction band minimum. Post-thermal annealing tuned the electron density of the 2D-ITO channel and enabled it to produce enhancement and depletion-mode 2D-ITO-TFTs. A full signal swing zero-VGS-load n-type metal-oxide semiconductor (NMOS) inverter composed of depletion-load/enhancement-driver 2D-ITO-TFTs and a complementary inverter with p-channel 2D-SnO-TFT were successfully demonstrated using all 2D-oxide-TFTs.

Atomically Thin Tin Monoxide-Based p-Channel Thin-Film Transistor and a Low-Power Complementary Inverter.

Atomically thin oxide semiconductors are significantly expected for next-generation cost-effective, energy-efficient electronics. A high-performance p-channel oxide thin-film transistor (TFT) was developed using an atomically thin p-type tin monoxide, SnO channel with a thickness of ∼1 nm, which was grown by a vacuum-free, solvent-free, metal-liquid printing process at low temperatures, as low as 250 °C in an ambient atmosphere. By performing oxygen-vacancy defect termination for the bulk-channel and back-channel surface of the ultrathin SnO channel, the presented p-channel SnO TFT exhibited good device performances with a reasonable TFT mobility of ∼0.47 cm2 V-1 s-1, a high on/off current ratio of ∼106, low off current of <10-12 A, and a subthreshold swing of ∼2.5 V decade-1, which was improved compared with the conventional p-channel SnO TFTs. We also fabricated metal-liquid printing-based n-channel oxide TFTs such as n-channel SnO2 and In2O3-TFTs and developed ultrathin-channel oxide-TFT-based low-power complementary inverter circuits with the developed p-channel SnO TFTs. The full swing of voltage-transfer characteristics with a voltage gain of ∼10 and a power dissipation of <4 nW for p-SnO/n-SnO2 and ∼120 and <2 nW for p-SnO/n-In2O3-CMOS inverters were successfully demonstrated.

Artificial Synapse Based on a 2D-SnO2 Memtransistor with Dynamically Tunable Analog Switching for Neuromorphic Computing.

A new type of two-dimensional (2D) SnO2 semiconductor-based gate-tunable memristor, that is, a memtransistor, an integrated device of a memristor and a transistor, was demonstrated to advance next-generation neuromorphic computing technology. The polycrystalline 2D-SnO2 memristors derived from a low-temperature and vacuum-free liquid metal process offer several interesting resistive switching properties such as excellent digital/analog resistive switching, multistate storage, and gate-tunability function of resistance switching states. Significantly, the gate tunability function that is not achievable in conventional two-terminal memristors provides the capability to implement heterosynaptic analog switching by regulating gate bias for enabling complex neuromorphic learning. We successfully demonstrated that the gate-tunable synaptic device dynamically modulated the analog switching behavior with good linearity and an improved conductance change ratio for high recognition accuracy learning. The presented gate-tunable 2D-oxide memtransistor will advance neuromorphic device technology and open up new opportunities to design learning schemes with an extra degree of freedom.

Back-Channel Defect Termination by Sulfur for p-Channel Cu2O Thin-Film Transistors.

The absence of a high-performance p-channel oxide thin-film transistor (TFT) is the major challenge faced in the current oxide semiconductor device technology. Simple solution-based back-channel subgap defect termination using sulfur was developed for p-channel cuprous oxide (Cu2O)-TFTs. We investigated the origin of poor device characteristics in conventional Cu2O-TFTs and clarified that it was mainly because of a back-channel donor-like defect of ∼2.8 ×1013 cm-2 eV-1, which originated from the interstitial Cu defect. Sulfur ion treatment using thiourea effectively reduced the back-channel defect down to <3 × 1012 cm-2 eV-1 and demonstrated the Cu2O-TFT with a saturation mobility of 1.38 ± 0.7 cm2 V-1 s-1, a s-value of 2.35 ± 1.22 V decade-1, and an on/off current ratio of ∼4.1 × 106. The improvement of device characteristics was attributed to the reduction of back-channel defect by the formation of a thin CuSO4 back-channel passivation layer by the chemical reaction of interstitial Cu with S and O ions. An oxide-based complementary inverter using a p-channel Cu2O-TFT and a n-channel a-In-Ga-Zn-O-TFT was demonstrated with a high voltage gain of ∼230 at VDD = 70 V.

Experimental demonstration of single-mode topological valley-Hall lasing at telecommunication wavelength controlled by the degree of asymmetry.

Topology plays a fundamental role in contemporary physics and enables new information processing schemes and wave device physics with built-in robustness. However, the creation of photonic topological phases usually requires complex geometries that limit the prospect for miniaturization and integration and dispossess designers of additional degrees of freedom needed to control topological modes on-chip. By controlling the degree of asymmetry (DoA) in a photonic crystal with broken inversion symmetry, we report single-mode lasing of valley-Hall ring cavities at telecommunication wavelength. The DoA governs four photon confinement regimes at the interface of topologically distinct valley-Hall domains and evidences an interplay between the width of the topological bandgap and the quality factor of ring-like modes for single-mode operation. Our results open the door to novel optoelectronic devices and systems based on compact topological integrated circuits.

Resistive switching of Sn-doped In2O3/HfO2 core-shell nanowire: geometry architecture engineering for nonvolatile memory.

Core-shell NWs offer an innovative approach to achieve nanoscale metal-insulator-metal (MIM) heterostructures along the wire radial direction, realizing three-dimensional geometry architecture rather than planar type thin film devices. This work demonstrated the tunable resistive switching characteristics of ITO/HfO2 core-shell nanowires with controllable shell thicknesses by the atomic layer deposition (ALD) process for the first time. Compared to planar HfO2 thin film device configuration, ITO/HfO2 core-shell nanowire shows a prominent resistive memory behavior, including lower power consumption with a smaller SET voltage of ∼0.6 V and better switching voltage uniformity with variations (standard deviation(σ)/mean value (µ)) of VSET and VRESET from 0.38 to 0.14 and from 0.33 to 0.05 for ITO/HfO2 core-shell nanowire and planar HfO2 thin film, respectively. In addition, endurance over 103 cycles resulting from the local electric field enhancement can be achieved, which is attributed to geometry architecture engineering. The concept of geometry architecture engineering provides a promising strategy to modify the electric-field distribution for solving the non-uniformity issue of future RRAM.

Self-Selecting Resistive Switching Scheme Using TiO2 Nanorod Arrays.

In this study, the resistive switching scheme using TiO2 nanorod arrays synthesized by a large-scale and low-cost hydrothermal process was reported. Especially, the nonlinear I-V characteristics of TiO2 nanorod arrays with a nonlinearity of up to ~10, which suppress the leakage current less than 10-4 Acm-2, were demonstrated, exhibiting a self-selecting resistive switching behavior. It provides a simple pathway for integration of RRAM crossbar arrays without additional stacking of active devices. The mechanisms of the nonlinear resistive switching behaviors were discussed in detail. In addition, the maximum array numbers of 79 for self-selecting RRAM cells were estimated. The results demonstrate an opportunity of using the concept of self-selecting resistive switching characteristics in a single material, which offers a new strategy to tackle the sneak path issue of RRAM in the crossbar arrays structure.

Tunable multilevel storage of complementary resistive switching on single-step formation of ZnO/ZnWO(x) bilayer structure via interfacial engineering.

Tunable multilevel storage of complementary resistive switching (CRS) on single-step formation of ZnO/ZnWOx bilayer structure via interfacial engineering was demonstrated for the first time. In addition, the performance of the ZnO/ZnWOx-based CRS device with the voltage- and current-sweep modes was demonstrated and investigated in detail. The resistance switching behaviors of the ZnO/ZnWOx bilayer ReRAM with adjustable RESET-stop voltages was explained using an electrochemical redox reaction model whose electron-hopping activation energies of 28, 40, and 133 meV can be obtained from Arrhenius equation at RESET-stop voltages of 1.0, 1.3, and 1.5 V, respectively. In the case of the voltage-sweep operation on the ZnO-based CRS device, the maximum array numbers (N) of 9, 15, and 31 at RESET-stop voltages of 1.4, 1.5, and 1.6 V were estimated, while the maximum array numbers increase into 47, 63, and 105 at RESET-stop voltages of 2.0, 2.2, and 2.4 V, operated by the current-sweep mode, respectively. In addition, the endurance tests show a very stable multilevel operation at each RESET-stop voltage under the current-sweep mode.

Single CuO(x) nanowire memristor: forming-free resistive switching behavior.

CuOx nanowires were synthesized by a low-cost and large-scale electrochemical process with AAO membranes at room temperature and its resistive switching has been demonstrated. The switching characteristic exhibits forming-free and low electric-field switching operation due to coexistence of significant amount of defects and Cu nanocrystals in the partially oxidized nanowires. The detailed resistive switching characteristics of CuOx nanowire systems have been investigated and possible switching mechanisms are systematically proposed based on the microstructural and chemical analysis via transmission electron microscopy.

Single-step formation of ZnO/ZnWO(x) bilayer structure via interfacial engineering for high performance and low energy consumption resistive memory with controllable high resistance states.

A spontaneously formed ZnO/ZnWOx bilayer resistive memory via an interfacial engineering by one-step sputtering process with controllable high resistance states was demonstrated. The detailed formation mechanism and microstructure of the ZnWOx layer was explored by X-ray photoemission spectroscopy (XPS) and transmission electron microscope in detail. The reduced trapping depths from 0.46 to 0.29 eV were found after formation of ZnWOx layer, resulting in an asymmetric I-V behavior. In particular, the reduction of compliance current significantly reduces the switching current to reach the stable operation of device, enabling less energy consumption. Furthermore, we demonstrated an excellent performance of the complementary resistive switching (CRS) based on the ZnO/ZnWOx bilayer structure with DC endurance >200 cycles for a possible application in three-dimensional multilayer stacking.

Manipulated transformation of filamentary and homogeneous resistive switching on ZnO thin film memristor with controllable multistate.

A bias polarity-manipulated transformation from filamentary to homogeneous resistive switching was demonstrated on a Pt/ZnO thin film/Pt device. Two types of switching behaviors, exhibiting different resistive switching characteristics and memory performances were investigated in detail. The detailed transformation mechanisms are systematically proposed. By controlling different compliance currents and RESET-stop voltages, controllable multistate resistances in low resistance states and a high resistance states in the ZnO thin film metal-insulator-metal structure under the homogeneous resistive switching were demonstrated. We believe that findings would open up opportunities to explore the resistive switching mechanisms and performance memristor with multistate storage.

ZnO1-x nanorod arrays/ZnO thin film bilayer structure: from homojunction diode and high-performance memristor to complementary 1D1R application.

We present a ZnO(1-x) nanorod array (NR)/ZnO thin film (TF) bilayer structure synthesized at a low temperature, exhibiting a uniquely rectifying characteristic as a homojunction diode and a resistive switching behavior as memory at different biases. The homojunction diode is due to asymmetric Schottky barriers at interfaces of the Pt/ZnO NRs and the ZnO TF/Pt, respectively. The ZnO(1-x) NRs/ZnO TF bilayer structure also shows an excellent resistive switching behavior, including a reduced operation power and enhanced performances resulting from supplements of confined oxygen vacancies by the ZnO(1-x) NRs for rupture and recovery of conducting filaments inside the ZnO TF layer. A hydrophobic behavior with a contact angle of ~125° can be found on the ZnO(1-x) NRs/ZnO TF bilayer structure, demonstrating a self-cleaning effect. Finally, a successful demonstration of complementary 1D1R configurations can be achieved by simply connecting two identical devices back to back in series, realizing the possibility of a low-temperature all-ZnO-based memory system.

HypertenGene: extracting key hypertension genes from biomedical literature with position and automatically-generated template features.

BACKGROUND: The genetic factors leading to hypertension have been extensively studied, and large numbers of research papers have been published on the subject. One of hypertension researchers' primary research tasks is to locate key hypertension-related genes in abstracts. However, gathering such information with existing tools is not easy: (1) Searching for articles often returns far too many hits to browse through. (2) The search results do not highlight the hypertension-related genes discovered in the abstract. (3) Even though some text mining services mark up gene names in the abstract, the key genes investigated in a paper are still not distinguished from other genes. To facilitate the information gathering process for hypertension researchers, one solution would be to extract the key hypertension-related genes in each abstract. Three major tasks are involved in the construction of this system: (1) gene and hypertension named entity recognition, (2) section categorization, and (3) gene-hypertension relation extraction. RESULTS: We first compare the retrieval performance achieved by individually adding template features and position features to the baseline system. Then, the combination of both is examined. We found that using position features can almost double the original AUC score (0.8140 vs.0.4936) of the baseline system. However, adding template features only results in marginal improvement (0.0197). Including both improves AUC to 0.8184, indicating that these two sets of features are complementary, and do not have overlapping effects. We then examine the performance in a different domain--diabetes, and the result shows a satisfactory AUC of 0.83. CONCLUSION: Our approach successfully exploits template features to recognize true hypertension-related gene mentions and position features to distinguish key genes from other related genes. Templates are automatically generated and checked by biologists to minimize labor costs. Our approach integrates the advantages of machine learning models and pattern matching. To the best of our knowledge, this the first systematic study of extracting hypertension-related genes and the first attempt to create a hypertension-gene relation corpus based on the GAD database. Furthermore, our paper proposes and tests novel features for extracting key hypertension genes, such as relative position, section, and template features, which could also be applied to key-gene extraction for other diseases.

PubMed-EX: a web browser extension to enhance PubMed search with text mining features.

UNLABELLED: PubMed-EX is a browser extension that marks up PubMed search results with additional text-mining information. PubMed-EX's page mark-up, which includes section categorization and gene/disease and relation mark-up, can help researchers to quickly focus on key terms and provide additional information on them. All text processing is performed server-side, freeing up user resources. AVAILABILITY: PubMed-EX is freely available at http://bws.iis.sinica.edu.tw/PubMed-EX and http://iisr.cse.yzu.edu.tw:8000/PubMed-EX/.

Semi-automatic conversion of BioProp semantic annotation to PASBio annotation.

BACKGROUND: Semantic role labeling (SRL) is an important text analysis technique. In SRL, sentences are represented by one or more predicate-argument structures (PAS). Each PAS is composed of a predicate (verb) and several arguments (noun phrases, adverbial phrases, etc.) with different semantic roles, including main arguments (agent or patient) as well as adjunct arguments (time, manner, or location). PropBank is the most widely used PAS corpus and annotation format in the newswire domain. In the biomedical field, however, more detailed and restrictive PAS annotation formats such as PASBio are popular. Unfortunately, due to the lack of an annotated PASBio corpus, no publicly available machine-learning (ML) based SRL systems based on PASBio have been developed. In previous work, we constructed a biomedical corpus based on the PropBank standard called BioProp, on which we developed an ML-based SRL system, BIOSMILE. In this paper, we aim to build a system to convert BIOSMILE's BioProp annotation output to PASBio annotation. Our system consists of BIOSMILE in combination with a BioProp-PASBio rule-based converter, and an additional semi-automatic rule generator. RESULTS: Our first experiment evaluated our rule-based converter's performance independently from BIOSMILE performance. The converter achieved an F-score of 85.29%. The second experiment evaluated combined system (BIOSMILE + rule-based converter). The system achieved an F-score of 69.08% for PASBio's 29 verbs. CONCLUSION: Our approach allows PAS conversion between BioProp and PASBio annotation using BIOSMILE alongside our newly developed semi-automatic rule generator and rule-based converter. Our system can match the performance of other state-of-the-art domain-specific ML-based SRL systems and can be easily customized for PASBio application development.

BIOSMILE web search: a web application for annotating biomedical entities and relations.

BIOSMILE web search (BWS), a web-based NCBI-PubMed search application, which can analyze articles for selected biomedical verbs and give users relational information, such as subject, object, location, manner, time, etc. After receiving keyword query input, BWS retrieves matching PubMed abstracts and lists them along with snippets by order of relevancy to protein-protein interaction. Users can then select articles for further analysis, and BWS will find and mark up biomedical relations in the text. The analysis results can be viewed in the abstract text or in table form. To date, BWS has been field tested by over 30 biologists and questionnaires have shown that subjects are highly satisfied with its capabilities and usability. BWS is accessible free of charge at http://bioservices.cse.yzu.edu.tw/BWS.