Solid-State Oxide-Ion Synaptic Transistor for Neuromorphic Computing

TL;DR

The paper tackles the challenges of neuromorphic hardware by introducing an all-solid-state oxide-ion synaptic transistor using BICUVOX as the oxide-ion conductor and LSF50 as the variable-resistance channel, compatible with conventional electronics. It demonstrates essential synaptic behaviors including long- and short-term plasticity, paired-pulse facilitation, and post-tetanic potentiation, with metrics such as low-energy per pulse, high endurance, and good linearity/symmetry. An ANN-like MNIST evaluation reports 96% accuracy, illustrating effective integration of the device into neural network analogs. Collectively, the work highlights oxide-ion–based iontronics as a viable path toward CMOS-compatible, low-power analog neuromorphic computing with deterministic ion intercalation mechanisms.

Abstract

Neuromorphic hardware facilitates rapid and energy-efficient training and operation of neural network models for artificial intelligence. However, existing analog in-memory computing devices, like memristors, continue to face significant challenges that impede their commercialization. These challenges include high variability due to their stochastic nature. Microfabricated electrochemical synapses offer a promising approach by functioning as an analog programmable resistor based on deterministic ion-insertion mechanisms. Here, we developed an all-solid-state oxide-ion synaptic transistor employing $\text{Bi}_2\text{V}_{0.9}\text{Cu}_{0.1}\text{O}_{5.35}$ as a superior oxide-ion conductor electrolyte and $\text{La}_\text{0.5}\text{Sr}_\text{0.5}\text{F}\text{O}_\text{3-$δ$}$ as a variable resistance channel able to efficiently operate at temperatures compatible with conventional electronics. Our transistor exhibits essential synaptic behaviors such as long- and short-term potentiation, paired-pulse facilitation, and post-tetanic potentiation, mimicking fundamental properties of biological neural networks. Key criteria for efficient neuromorphic computing are satisfied, including excellent linear and symmetric synaptic plasticity, low energy consumption per programming pulse, and high endurance with minimal cycle-to-cycle variation. Integrated into an artificial neural network (ANN) simulation for handwritten digit recognition, the presented synaptic transistor achieved a 96% accuracy on the MNIST dataset, illustrating the effective implementation of our device in ANNs. These findings demonstrate the potential of oxide-ion based synaptic transistors for effective implementation in analog neuromorphic computing based on iontronics.

Figures (27)

• Figure 1: Scanning XRD of BICUVOX thin film on single crystal LSAT (red) and Si$\vert$STO-substrate (blue) before microfabrication of the synaptic transistors shows a highly epitaxial orientation in the [00l] plane conforming the great quality of produced thin films. Diamond-indicated peak is Si$_\text{100}$-substrate.
• Figure 2: Synaptic transistor microfabrication route. See details in methods section.
• Figure 3: Device Architecture. A) Three microfabricated synaptic transistor devices on the same LSAT-BICUVOX-LSF50 substrate with 2;3;5 separation between channel and reservoir respectively. B) SEM zoom of reported synaptic transistors (3 gap). C) calculation channel dimensions via thickness of deposited LSF50 thin film on BICUVOX before microfabrication (35 thickness).
• Figure 4: AFM images of layered BICUVOX surface before microfabrication confirm XRD results showing highly uniform and epitaxial orientation of the BICUVOX thin film. a) Layered, epitaxial BICUVOX surface. b) LSF50 surface showing great compatibility of LSF50 with epitaxial BICUVOX. c) 3D AFM image of channel and reservoir dimensions. 2px mean value filter was applied. d) AFM height profile of channel and reservoir after microfabrication. It is clear that during the ion-milling, the BICUVOX in the gap between channel and reservoir has been $\approx\qty{25}{\nano\metre}$ been etched into, leaving $\approx\qty{115}{\nano\metre}$ of BICUVOX thin film between channel and reservoir for our synaptic transistor.
• Figure 5: EIS of the BICUVOX thin film after microfabrication measured between channel and reservoir electrode referring to Figure 1b showing great ionic conduction assigned to oxide ions.