Interface engineering in ZnO/CdO hybrid nanocomposites to enhanced resistive switching memory for neuromorphic computing

• DOI: 10.1016/j.jcis.2023.12.084
• Publication Date: 2023-12-21T01:45:36Z
• Authors: Faisal Ghafoor; Honggyun Kim; Bilal Ghafoor; Shania Rehman; Muhammad Asghar Khan; Jamal Aziz; Muhammad Rabeel; Muhammad Faheem M...; Ghulam Dastgeer; Myoung-Jae Lee; Muhammad Farooq Khan; Deok-kee Kim
• Keywords: Zinc oxide (ZnO); Oxygen vacancies (Vo); ROOM-TEMPERATURE; RATIO; Cadmium oxide (CdO); Nanocomposite (NC); ZNO; Conductive filament (CF); OPTICAL-PROPERTIES; REDUCED GRAPHENE OXIDE; 530; 620

Resistive random-access memory (RRAMs) has attracted significant interest for their potential applications in embedded storage and neuromorphic computing. Materials based on metal chalcogenides have emerged as promising candidates for the fulfilment of these requirements. Due to its ability to manipulate electronic states and control trap states through controlled compositional dynamics, metal chalcogenide RRAM has excellent non-volatile resistive memory properties. In the present we have synthesized ZnO-CdO hybrid nanocomposite by using hydrothermal method as an active layer. The Ag/C15ZO/Pt hybrid nanocomposite structure memristors showed electrical properties similar to biological synapses. The device exhibited remarkably stable resistive switching properties that have a low SET/RESET (0.41/-0.2) voltage, a high RON/OFF ratio of approximately 105, a high retention stability, excellent endurance reliability up to 104 cycles and multilevel device storage performance by controlling the compliance current. Furthermore, they exhibited an impressive performance in terms of emulating biological synaptic functions, which include long-term potentiation (LTP), long-term depression (LTD), and paired-pulse facilitation (PPF), via the continuous modulation of conductance. The hybrid nanocomposite memristors notably achieved an impressive recognition accuracy of up to 92.6 % for handwritten digit recognition under artificial neural network (ANN). This study shows that hybrid-nanocomposite memristor performance could lead to efficient future neuromorphic architectures.