Published May 2024 | Version v1
Journal article

Inkjet-printed tungsten oxide memristor displaying non-volatile memory and neuromorphic properties

  • 1. Institute of Nanotechnology (INT), Karlsruhe Institute of Technology (KIT), Eggenstein-Leopoldshafen, 76344 (Germany)
  • 2. Laboratory for Electron Microscopy (LEM), Microscopy of Nanoscale Structures and Mechanisms (MNM), Karlsruhe Institute of Technology KIT, Karlsruhe, 76131 (Germany)
  • 3. Nanostructure Service Laboratory (NSL), Karlsruhe Institute of Technology (KIT), Karlsruhe, 76131 (Germany)
  • 4. Karlsruhe Nano Micro Facility (KNMF), Karlsruhe Institute of Technology (KIT), Eggenstein-Leopoldshafen, 76344 (Germany)

Description

Printed electronics including large-area sensing, wearables, and bioelectronic systems are often limited to simple circuits and hence it remains a major challenge to efficiently store data and perform computational tasks. Memristors can be considered as ideal candidates for both purposes. Herein, an inkjet-printed memristor is demonstrated, which can serve as a digital information storage device, or as an artificial synapse for neuromorphic circuits. This is achieved by suitable manipulation of the ion species in the active layer of the device. For digital-type memristor operation resistive switching is dominated by cation movement after an initial electroforming step. It allows the device to be utilized as non-volatile digital memristor, which offers high endurance over 12 672 switching cycles and high uniformity at low operating voltages. To use the device as an electroforming-free, interface-based, analog-type memristor, anion migration is exploited which leads to volatile resistive switching. An important figure of merits such as short-term plasticity with close to biological synapse timescales is demonstrated, for facilitation (10-177 ms), augmentation (10s), and potentiation (35 s). Furthermore, the device is thoroughly studied regarding its metaplasticity for memory formation. Last but not least, the inkjet-printed artificial synapse shows non-linear signal integration and low-frequency filtering capabilities, which renders it a good candidate for neuromorphic computing architectures, such as reservoir computing. (© 2023 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH)

Additional details

Identifiers

Publishing Information

Journal Title
Advanced Functional Materials (Internet)
Journal Volume
34
Journal Issue
20
Journal Page Range
p. 1-12
ISSN
1616-3028
CODEN
AFMDC6

Optional Information

Notes
AID: 2302290; 3D matter made to order