Pham Phu Quan, a master's student in Material Science from the K32(2022), successfully defended his thesis titled: Design and Fabrication of Self-Rectifying Memristors Based on CrOx/TiOy for Artificial Synapse Applications.
The thesis introduces several notable innovations in the development of a self-rectifying memristor device. It focuses on fabricating a stacked CrOx/TiOy thin-film structure that exhibits excellent endurance in resistive switching, driven by ion migration between the layers, as confirmed through mathematical modeling. The fabrication processes and material properties were optimized to achieve high performance, enabling the device to operate at voltages below 5 V, sustain durability over 1,000 switching cycles, and attain a rectifying ratio exceeding 25. A computational model was developed for the memristor, allowing its behavior to be simulated using software tools such as LT Spice, NG Spice, and Xschem. The memristor also demonstrated the ability to replicate biological synaptic functions, such as potentiation and depression, making it suitable for artificial neural networks and brain-inspired computing. Its capability to update weights like biological synapses highlights its potential in advanced AI systems. Furthermore, the fabrication process was successfully scaled from a single-cell device to an integrated 256-cell memristor chip, marking a significant step toward practical applications.
This innovative technology offers promising applicability in several areas. The memristor's self-rectifying behavior and multi-level data storage capabilities make it ideal for high-density memory chips, providing an efficient and compact solution for synaptic weight storage in computing systems. Its analog resistive switching behavior supports in-memory computing, enabling simultaneous data processing and storage, which reduces energy consumption and increases processing speed. The successful demonstration of a 256-cell memristor chip positions this technology as a key enabler for artificial synapse chips, essential for cognitive computing and applications like pattern recognition, paving the way for advances in AI and brain-inspired technologies.
Funding
This work was supported by the Vingroup Innovative Fund under grant number VINIF.2023. DA130.
The authors express their sincere thanks to the crew of the Center for INOMAR, VNU-HCM for their continuous support and help in using the XRD and AFM facilities. Addition,
The authors express their sincere thanks to the crew of the Microsystem Laboratory, School of Engineering, Swiss Federal Institute of Technology Lausanne, Switzerland for their continuous support and help in fabricating and evaluating the stencils.
Post a Comment