Exploring the potential of 2D PtTe2-based memristors for neuromorphic computing

Abstract

Neuromimetic devices have emerged as transformative technologies with the potential to redefine traditional computing paradigms and enable advanced artificial neural systems. Among various innovative materials, two-dimensional (2D) materials have garnered attention as frontrunners for next-generation device fabrication. In this work, we report the fabrication and comprehensive characterization of a memristor based on 2D PtTe2. The device demonstrates exceptional performance metrics, including a high OFF/ON ratio, low switching voltage, and long data retention time. Leveraging density functional theory calculations, we unravel the underlying conduction mechanism, revealing the pivotal role of Ag conductive filaments in resistive switching behavior. Furthermore, the neuromorphic capabilities of the PtTe2 memristor were evaluated through its emulation of key brain-inspired synaptic functionalities, such as long-term depression/enhancement, paired-pulse facilitation, and spike-timing-dependent plasticity. By modulating its electrical conductance, we implemented a convolutional neural network for MNIST handwritten digit recognition, achieving a remarkable accuracy of 97.49%. To further illustrate its adaptive learning capabilities, we demonstrated a Pavlov's dog experiment using the device. This study establishes 2D PtTe2 as a promising material for neuromorphic applications and represents a critical step forward in bridging the gap between advanced materials and next-generation computing architectures. These findings lay a robust foundation for future exploration of PtTe2 in the field of neuromorphic engineering.