Inkjet‐Printed Tungsten Oxide Memristor Displaying Non‐Volatile Memory and Neuromorphic Properties

Hongrong Hu,Alexander Scholz,Christian Dolle,Alexander Zintler,Aina Quintilla,Yan Liu,Yushu Tang,Ben Breitung,Gabriel Cadilha Marques,Yolita M. Eggeler,Jasmin Aghassi‐Hagmann

DOI: https://doi.org/10.1002/adfm.202302290

IF: 19

2023-07-08

Advanced Functional Materials

Abstract:An inkjet‐printed tungsten oxide memristor is developed, which exhibits either digital‐type switching for data storage or analog‐type memristor behavior for neuromorphic circuits. The distinction between both operation modes is determined by either applying an electroforming voltage or excluding it. The devices show high endurance, excellent synaptic properties with respect to short‐term plasticity, metaplasticity, and exhibit non‐linear signal integration and filtering. 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.

materials science, multidisciplinary,chemistry, physical,physics, applied, condensed matter,nanoscience & nanotechnology