Highly Uniform Silver Ion Memristors With Ultralow Leakage Current for Constructing Homogeneously Spiking LIF Neurons

The stable and threshold-friendly electrochemical-metallization-based threshold switching (TS) memristors are promising candidates to build artificial leaky integrate-and-fire (LIF) neurons for neuromorphic computing. In this work, Ag/Al2O3/HfO2/TiN memristors have been fabricated with small <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${V} _{\text {th}}$ </tex-math></inline-formula>/<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${V} _{\text {h}}$ </tex-math></inline-formula> variation coefficients of 0.021 and 0.051, a large selectivity (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$10^{{6}}$ </tex-math></inline-formula>), and ultralow leakage current (100 fA). The excellent volatile properties can be attributed to the insertion of a 2-nm Al2O3 buffer layer, and the high uniformity of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${V} _{\text {th}}$ </tex-math></inline-formula>/<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${V} _{\text {h}}$ </tex-math></inline-formula> causes homogeneous spiking behavior for LIF neurons. Moreover, spiking neural network (SNN) based on the memristor array is fabricated for image classification with a recognition accuracy of 93.2%, which provides guidelines for constructing stable and high-precision SNN systems by Ag+ ion-based artificial neurons.