Interfacial Ion-Trapping Electrolyte-Gated Transistors for High-Fidelity Neuromorphic Computing

Abstract

Li⁺ electrolyte-gated transistors (EGTs) have received much attention as artificial synapses for neuromorphic computing. EGTs, however, have been still challenging to achieve long-term synaptic plasticity, which should be linearly and symmetrically controlled with the magnitude of electrical potential at the gate electrode. Herein, a fluoroalkylsilane (FAS) self-assembled monolayer (SAM) is introduced as a channel-electrolyte interlayer with the function of sequential ion-trapping in Li⁺ EGTs. It is demonstrated that the retention of Li⁺ ions can be enhanced, resulting in stable non-volatile channel conductance update with high fidelity, linearity, and symmetry in EGTs treated with FAS with 5 fluoroalkyl chains. Through investigating electrical analysis and chemical analysis, it is verified that fluoroalkyl chains enable the sequential ion-trapping at the channel-electrolyte interface by coulombic attraction between Li⁺ ions and fluorocarbons. Simulations of artificial neural networks using 20 × 20 digits show FAS-treated EGTs are suitable as artificial synapses with an accuracy of 89.71% by identical gate pulses and 91.97% by non-identical gate pulses. A methodological approach is newly introduced for developing synaptic devices based on EGTs for neuromorphic computing with high fidelity.