The Influence of Conducting Cation Selections on the Electrolyte-Gated Synaptic Transistors—Electrochemical and Mechanical Analysis

Neuromorphic architecture is a protocol that emulates the brain's perception, learning, and information processing using parallel computing pipelines. In contrast to Von Neumann computation, this architecture is characterized by fast computation and power-efficient performance, making them effective for big data processing in artificial intelligence. Among the various devices developed to implement the architecture to date, electrolyte-gated synaptic transistors (EGTs) have attracted much attention due to the advantage of precise conductivity control, fast response time and low operating voltage via ion-assisted signal transmission. Their three-terminal design avoids issues like sneak paths and signal overlap seen in two-terminal devices. However, understanding the physicochemical factors and mechanisms of EGTs related to ion behavior remains limited, despite ions exerting a critical influence on synaptic performance in EGTs.<br/>To address this, we propose a novel analytical method with electrochemical characterization using three alkali cations, Li+, Na+ and K+, and poly (ethylene oxide) in EGTs (AGTs). While they have similar chemical properties to monovalent ions, the ionic sizes vary significantly: 76 pm for Li+, 102 pm for Na+, and 138 pm for K+. We used cyclic voltammetry (CV) and X-ray photoelectron spectroscopy (XPS) to investigate the non-volatile memory mechanisms of EGTs. Additionally, electrochemical impedance spectroscopy (EIS), scanning electron microscopy (SEM), differential scanning calorimetry (DSC), and X-ray diffraction analysis (XRD) explored how electrolyte properties and ion dynamics influence the analog conductivity of EGTs.<br/>The result shows that b-values extracted by CV measurements were 0.55, 0.63, and 0.77 for Li+, Na+, and K+, respectively, indicating the dominant reaction switches from redox reaction to electrical double layer capacitance (EDLC) behavior as the ion size increases. In turn, Li-AGTs exhibited the best non-volatile memory, retaining 34.2% of the initial current after 100 s, while K-AGTs returned to the initial current of 1 µA due to ionic self-diffusion derived from EDLC. EIS measurements revealed a decrease in ionic conductivity with increasing ionic radius, with K-PEO having the lowest value of 4.65 × 10-7 mScm^-1. The XRD and SEM results showed that the larger ions have a lower solubility in the electrolyte, forming ionic entanglement and increasing crystallinity in K-PEO. The activation energies (Ea) for Li+, Na+, and K+ were 0.39, 0.67, and 1.66 eV, respectively. The crystallized K-PEO mitigated rapid ionic diffusion/drift transport, resulting in near-ideal nonlinearity values of 2.27 and -2.85 in the potentiation and depression regions. This systematic approach with electrochemical analysis facilitates intuitive interpretation, advancing the development of high-performance artificial synaptic EGTs.