An ion-gating synaptic memristor based on tri-layer HfOx composition regulation†
Abstract
In this work, we developed tri-layer HfOx/HfO2/HfOx memristors that exhibit high consistency and good linearity, making them suitable for high-efficiency neuromorphic computing. The HfO2 intermediate layer serves as an ion-gating layer, enabling the precise localization and shaping of conductive pathways while regulating oxygen vacancy (VO) migration. By optimizing the VO difference Δ (Δ = 2 − x) and the ion-gating HfO2 interlayer, we were able to precisely control the formation and rupture of conductive filaments (CFs) within the HfO2 interlayer, leading to improved consistency, linearity and continuity of resistance variation. Notably, the HfO1.7/HfO2/HfO1.7 (T-HfO1.7) device demonstrated the highest low resistance consistency (1.7%) for memory function. Furthermore, this device exhibited essential synaptic functions, including long-term potentiation (LTP), paired-pulse facilitation (PPF), and spike-timing-dependent plasticity (STDP). The conductance modulation process of T-HfO1.7 achieves high linearity (αLTP = 1.55). Moreover, a Hopfield Neural Network (HNN) constructed using this device achieved a high image recognition accuracy of 95.6%. This work introduces a straightforward approach to improve the consistency and linearity of memristive behavior, paving the way for enhanced performance in neuromorphic applications.