Nanolayer Materials for Electrochemical Ionic Synapses in Neuromorphic Computing

Interface and Electronic Materials Laboratory

Deep Learning has revolutionised various fields, from computer vision to natural language processing. However, conventional digital deep learning hardware struggles to match the energy efficiency and processing speed of the human brain. Analog deep learning devices offer a promising alternative. This project focuses on harnessing the potential of nanolayer materials to create highly efficient and scalable analog deep learning devices. The project aims to investigate and develop novel nanolayer materials suitable for analog deep-learning thanks to their ionic conductivity. Explore the properties of these materials, including ion mobility, charge retention, and compatibility with device fabrication processes. The student will develop fabrication techniques for analog deep learning devices utilising nanolayer materials. This includes the design and manufacturing of analog synaptic devices, memristors, or other relevant components. In a subsequent step, such devices will be implemented into analog neural network architectures that leverage the unique properties of nanolayer materials. In this way we will explore the potential for in-memory computing, spike-based coding, and other analog computing paradigms. Of key importance is to evaluate the performance of analog deep learning devices in terms of energy efficiency, processing speed, and accuracy. This project requires collaboration across various disciplines, including materials science, electrical engineering, computer science, and machine learning. Collaborative efforts will help address the multifaceted challenges associated with developing nanolayer-based analog deep learning devices. This requires hands-on materials synthesis, device design and manufacture, electrical and optical measurements of materials, as well as data processing, analysis, and modelling of the observed characteristics. This project aims to advance the field of neuromorphic computing by exploring innovative materials and hardware architectures. The project has the potential to significantly improve energy efficiency and processing capabilities, opening up new possibilities for AI applications and addressing the growing demand for efficient, brain-inspired computing solutions.