Short-term memory characteristics of TiN/WOX/FTO-based transparent memory device

• Published in: Chinese journal of physics (Taipei) 2024-04, Vol.88, p.1044-1052
• Main Authors: Kim, Gyeongpyo, Cho, Youngboo, Kim, Sungjun
• Format: Article
• Language: English
• Subjects: Memristor; Neuromorphic; Reservoir computing; Resistive switching
• ISSN: 0577-9073
• DOI: 10.1016/j.cjph.2024.02.049

•Transparent memristor with TiN/WOX/FTO stack is proposed.•Leakage current is minimized by self-rectifying characteristics.•Linear control of short-term memory operations and device states is shown.•Neuromorphic system and reservoir computing are demonstrated. This work proposes a TiN/WOX/FTO device as a candidate for transparent artificial synapses. First, transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), and spectrophotometer analysis are used to analyze the material properties of the TiN/WOX/FTO device, whose optical transmittance exceeds 68 % for all wavelengths in the visible region. Specifically, the highest transparency of 78.1 % is observed at 580 nm. Self-rectifying switching characteristics are advantageous for suppressing leakage current in a crossbar array structure, which is achieved through a TiON layer due to the defects in an asymmetrical element stack and WOX/TiN interface; this shows that controlling the compliance current enables adjustment of the rectification ratio. Because of the self-rectifying characteristics, the improved 1/2 V scheme shows a better read margin. The programmed data show short-term characteristics that are lost over time, and short-term memory (STM) is confirmed through natural depression by changing the time interval of the read pulse. Based on these STM characteristics, biological synapse features such as potentiation, depression, and paired-pulse facilitation (PPF) are emulated. The typical potentiation and depression are also implemented by optimizing the set and reset pulses. Moreover, a high recognition accuracy (>94.68 %) is obtained for the modified National Institute of standards and Technology (MNIST) pattern. Finally, we demonstrate the feasibility of a reservoir computing device. [Display omitted]