Oxide‐Based Electrolyte‐Gated Transistors for Spatiotemporal Information Processing

Abstract

systems, which can be integrated with terminal sensors to form intelligent sensory systems. [3-7] Information processing in biological sensory nervous systems involves billions of neurons interconnected through trillions of synapses, constituting immense neural networks. [8] Compared with traditional von Neumannbased computing architecture, neural networks greatly reduce time-and energy consumption by taking the advantage of co-location of logic and memory, hyperconnectivity, robustness, and massively parallel processing. [9] As the third-generation artificial neural network, spiking neural networks (SNNs) are inspired by biological nervous systems. They employ spiking neurons as computational units that process information with timing of spikes. [10] Therefore, SNNs provide the potential for spatiotemporal information processing with high time-and energy-efficiency. The learning and recognition functions aiming at spatiotemporal patterns have been demonstrated in SNN formed by resistive switching synaptic devices. [11] However, this synaptic device suffers from several performance limitations such as the discrete weight update, write variation, and high energy programing related to the filamentary switching mechanism. [12-17] Recently, He et al. use capacitively coupled multi-terminal neurotransistors for spatiotemporal information processing, where the neurotransistors mimic the dendritic discriminability of different spatiotemporal input sequences. [18] To physically emulate neural networks in hardware, large-scale crossbar Spiking neural networks (SNNs) sharing large similarity with biological nervous systems are promising to process spatiotemporal information and can provide highly time-and energy-efficient computational paradigms for the Internet-of-Things and edge computing. Nonvolatile electrolyte-gated transistors (EGTs) provide prominent analog switching performance, the most critical feature of synaptic element, and have been recently demonstrated as a promising synaptic device. However, high performance, large-scale EGT arrays, and EGT application for spatiotemporal information processing in an SNN are yet to be demonstrated. Here, an oxide-based EGT employing amorphous Nb 2 O 5 and Li x SiO 2 is introduced as the channel and electrolyte gate materials, respectively, and integrated into a 32 × 32 EGT array. The engineered EGTs show a quasi-linear update, good endurance (10 6) and retention, a high switching speed of 100 ns, ultralow readout conductance (<100 nS), and ultralow areal switching energy density (20 fJ µm −2). The prominent analog switching performance is leveraged for hardware implementation of an SNN with the capability of spatiotemporal information processing, where spike sequences with different timings are able to be efficiently learned and recognized by the EGT array. Finally, this EGT-based spatiotemporal information processing is deployed to detect moving orientation in a tactile sensing system. These results provide an insight into oxide-based EGT devices for energy-efficient neuro...

In situ Parallel Training of Analog Neural Network Using Electrochemical Random-Access Memory

“…By encoding information more sparsely in the temporal domain and restricting the analog requirements for information transmission, SNNs are widely considered more likely to achieve the energy efficiency and error tolerance of biological computing system (Wang et al, 2020). Spike time-dependent plasticity (STDP) like functionality have been demonstrated in organic and inorganic ECRAM (van de Burgt et al, 2017;Sharbati et al, 2018;Li et al, 2020b). While STDP or STPD-like rules do not directly implement stochastic gradient descent backpropagation, they effectively sample an input and may approximate a statistical method known as expectation maximization (Nessler et al, 2013).…”

A Flexible Mott Synaptic Transistor for Nociceptor Simulation and Neuromorphic Computing

“…Realization of multi‐states in synaptic strength is a requisite condition for the artificial neural computing. [ 12,15,20 ] Here, it is realized with continuous positive (1.8 V, 2 s) and negative (−1.5 V, 2 s) gate pulse. As Figure 5d shows, one hundred and twenty (120) discrete conductance states (orange discrete points) can be obtained through fifty positive and seventy negative gate pulses in one cycle.…”

“…The high recognition accuracy of our synaptic transistors is ascribed to the high stability of devices, facilitating low non‐ideal factors, such as C2C variation, nonlinear parameter of potentiation and depression processes, symmetry nonlinearity factor between potentiation and depression. [ 12,20,58 ] The estimated C2C variation is shown in Figure S7, Supporting Information. The nonlinear parameter and symmetry nonlinearity factor are also evaluated by the formulas provided in experiment section, and the symmetry nonlinearity factor is 0.58, the nonlinear parameter of potentiation and depression processes are 0.026 and 0.045, respectively.…”

“…As the basic unit of the biological nervous system, synapse is a functional connection point of neurons which can modulate its connection strength (synaptic weight) to store and process information simultaneously. [ 7–9 ] To date, artificial synaptic devices including two‐terminal memristors [ 10–14 ] and three‐terminal transistors, [ 15–20 ] have achieved significant progress, and it should be mentioned that most works have remained at the devices on rigid substrate. With the development of flexible electronics, integrating and applying the synaptic devices in flexible electronic skins and soft robots with stable synaptic functions are in urgent demand.…”

Non‐Volatile Electrolyte‐Gated Transistors Based on Graphdiyne/MoS2 with Robust Stability for Low‐Power Neuromorphic Computing and Logic‐In‐Memory

“…[11] For the EDL-based EGTs, ions (e.g., Li + and H + ) in electrolyte accumulate at the channel/electrolyte interface and modulate the channel conductance via EDL gating effect, [25,33,34] whereas the ions can penetrate into the channel under a relatively high gate voltage and the channel conductance can be modulated by electrochemical doping or redox for EGTs based on electrochemical doping or redox. [26,27,29,[35][36][37][38] As a result, the EDL-based EGTs feature a large dynamic range (G max /G min ), a low operation voltage and a small leakage current, while the EGTs based on electrochemical doping and redox exhibit a long retention time since the intercalated ions can be trapped in the channel even after the gate pulse. However, as summarized in Table S1, Supporting Information, EGTs in previous reports always suffer some disadvantages, for example, poor retention characteristics for EDL-based EGTs and large operation voltage for EGTs based on electrochemical doping or redox, due to their device architectures and working principles.…”