Unveiling the role of grain boundaries in driving in-plane conductive filament formation in MoS2 neuromorphic devices

Abstract

The growing demand for energy-efficient artificial intelligence systems has driven the exploration of neuromorphic devices that mimic the human brain's synaptic behavior. Among various materials investigated for neuromorphic devices, two-dimensional transition metal dichalcogenides, particularly MoS2, exhibit promising properties for memristors and memtransistors due to their tunable bandgap, mechanical flexibility, and high surface activity. Although vertical conductive filament (CF) formation in MoS2-based devices has been well-documented, the mechanisms governing in-plane CF formation in MoS2 remain poorly understood, particularly in the presence of grain boundaries (GBs). In this work, we systematically investigate the role of 4|6 and 5|7 GBs in monolayer MoS2 on the in-plane formation and migration of copper-based CFs using first-principles density functional theory and molecular dynamics simulations. Our results reveal that the 4|6 GB significantly lowers the copper migration energy barrier (1.33 eV) compared to pristine (1.56 eV) and 5|7 GB-containing MoS2 (2.75 eV). Differential charge density analysis and band structure calculations confirm that GBs enhance binding affinity and modulate local electronic properties, promoting metallic behavior upon Cu incorporation. Molecular dynamics simulations under an applied electric field further reveal that Cu migration and CF formation preferentially occur along the 4|6 GB. These findings provide critical insights into defect-engineered CF modulation and highlight the potential of 4|6 GBs to enhance the performance and reliability of lateral MoS2 neuromorphic devices.