Resistive switching dynamics in MoSe2-ZnO nanoheterostructures for energy-efficient neuromorphic application

An artificial synapse is integral to neuromorphic computing, a field poised to overcome the limitations of the traditional von Neumann architecture. Memristors, with their tunable, non-volatile resistive switching (RS) states, hold significant promise for acting as artificial synapses, facilitating both data storage and processing within the same physical unit. In this study, we report on memristive devices based on a hydrothermally synthesized MoSe₂-ZnO nanoheterostructure, integrated between upper Ni/Ag and lower FTO electrodes, with a comprehensive investigation into their RS characteristics, synaptic functionalities, and potential for neuromorphic computing applications. The structural, compositional, and electronic properties of the MoSe₂-ZnO nanoheterostructure were probed using XRD, Raman spectroscopy, FESEM, HRTEM, EDS, and XPS analyses. The fabricated Ag/MoSe₂-ZnO/FTO memristor exhibited reliable analog resistive switching (ARS) behavior over a low operational voltage range (-1 V to +1 V). The device successfully emulated key synaptic functions, including potentiation and depression, under microsecond pulse stimuli (1 µs) at multiple read voltages (0.2–0.6 V), closely replicating biological synaptic plasticity. Additionally, assessments of endurance, data retention, device-to-device (D2D), and cycle-to-cycle (C2C) reliability confirmed consistent analog switching behavior and stable operational performance. A mechanistic analysis revealed a hybrid resistive switching mechanism, involving both Ag⁺-based conductive filament formation/dissolution and charge trapping/detrapping within the MoSe₂-ZnO matrix. This dual-mode conduction was supported by double-logarithmic I–V analysis and energy band diagram illustrations, clarifying the role of interface dynamics and barrier modulation under bias.