Endogenous Overdrive Stabilization in Monolithic One-Transistor-One-Electrolyte-Gated Transistor Devices for Mitigating Nonlinear Synaptic Updates

Abstract

Solid-state electrolyte-gated transistors (EGTs) enable synaptic state updates for neuromorphic computing. When integrated with a driving transistor, the effective programming voltage is expected, under an idealized low-impedance model, to decrease as the conductance of EGT increases, leading to nonlinear synaptic update. Here, we show that this assumption does not hold in a CMOS-compatible 1T1E architecture that monolithically integrates a silicon NMOS with a LiPON-based EGT. In situ measurements reveal the opposite trend, where the effective programming voltage increases during pulsed operation. This behavior arises from electrolyte-dominated impedance evolution, which drives voltage redistribution and compensates the evolving electrochemical barrier. This endogenous stabilization of the electrochemical overdrive mitigates nonlinear synaptic update, enabling improved linearity and stable multilevel conductance modulation. These results indicate a regime-dependent framework in which programming dynamics are governed by impedance hierarchy rather than fixed biasing conditions.