Electrolyte Gating WO₃ Synaptic Transistor with Crossbar Array Structure

Neuromorphic computing is gaining attention compared to conventional computing systems based on the von Neumann architecture, which have limitations in processing speed due to delays when massive amounts of data are transferred between the central processing unit (CPU) and memory. To overcome this von Neumann bottleneck, synaptic devices that mimic the human brain have become attractive in neuromorphic systems. Especially, many researchers have been interested in electrolyte-gating transistors (EGTs) due to their low switching voltage, enabling low-power consuming devices. Also, their non-volatile memory characteristics can emulate potentiation in synapse, synaptic behaviors such as synaptic plasticity and linearity. Synaptic plasticity allows the human brain to process the signals and store information simultaneously. Biological plasticity can be achieved by converting connection strength between synapses, called synaptic weight, in response to neural activities. Linearity can be confirmed from the shape of the long-term potentiation (LTP) and long-term depression (LTD) region. Effectively emulating the biological synaptic behaviors in electronic devices would help to meet the needs of fast and efficient neuromorphic learning devices.<br/>In this study, tungsten oxide and nafion were selected as a channel layer and electrolyte to implement EGTs, respectively. Tungsten oxide has a structural advantage because the vacant sites provide space for ion intercalation and shows electrochromic properties, coloring and bleaching by movement of ions between electrolyte and channel. When the ions in the electrolyte inserted into channel layer, coloration occurs. In contrast, when the ions extracted from the channel, bleaching occurs. This reaction is electrochemical and reversible reaction, which are related to memory characteristics. Nafion has a high chemical stability and excellent ionic conductivity. Compared to proton, many studies have used Li ion for EGTs but Li ion has a fatal disadvantage, which is not complementary metal–oxide–semiconductor (CMOS) compatible. In addition, most of the previous studies have been focused on interpreting the characteristics of unit device but it is not suitable for industrial applications requiring arrays. By arranging transistors in array, each device can be operated independently, suggesting possibility of data visualization. In case of proton conducting EGTs, there are several limitations to make arrays. First of all, there are only few options on materials and engineering for proton gating electrolyte. Also, considering array of nafion electrolyte, it requires a delicate process such as photolithography to implement array but it is difficult because nafion dissolves in acetone. Herein, we proposed nafion patterning method and EGT arrays. Due to the electrochromic properties of tungsten oxide, reversible reaction can be shown between tungsten oxide and proton cation, which can be extended to non-volatile memory characteristics and synaptic behaviors. Furthermore, three-terminal transistor array was arranged with nafion electrolyte patterning. Our research provides the framework for oxide semiconductor EGTs with polymer electrolyte, requiring further research for real application in brain-imitating neuromorphic system.