GCMO-based memristors for future neuromorphic processor

Abstract

This dissertation investigates resistive switching (RS) and neuromorphic functionality in memristor devices based on the mixed-valence perovskite manganite Gd0.2Ca0.8MnO3 (GCMO), with emphasis on the role of substrates and interfacial processes in device operation. Particular focus is placed on how structural quality, substrate-induced defects, and interfacial redox processes influence the switching characteristics of the devices and enable both volatile and non-volatile RS.

Thin-film GCMO layers were fabricated using pulsed laser deposition on different substrates, allowing control of crystallinity and defect density through substrate selection and growth conditions. Structural properties were characterized using X-ray diffraction, and the electrical behavior of the devices was studied through current-voltage measurements together with retention and endurance tests. Memristor devices with different active areas and electrode configurations were formed by depositing patterned metal electrodes on the GCMO films, creating an active AlO/GCMO interface responsible for resistive switching.

The results show that the switching behavior of GCMO memristors is influenced by the structural quality of the films and the defect landscape induced by the substrate. High-quality epitaxial films exhibit stable bipolar resistive switching, while variations in crystallinity modify the state retention through increased oxygen back-diffusion in polycrystalline thin films. Through X-ray photoelectron spectroscopy and area-dependent RS, the redox-active Al/GCMO interface is found to play a central role in the formation and modulation of resistive states.

In addition to synaptic functionality in crossbar-compatible devices, the results indicate neuronal-like behavior such as leaky-integrate dynamics. These findings highlight the potential of GCMO memristors as building blocks for future neuromorphic device architectures.