Energy-efficient Neuromorphic 2D Devices and Circuits for Edge AI Computing

ENERGIZE aims to implement novel neuromorphic hardware based on two-dimensional (2D) materials for energy efficient artificial intelligence (AI).

This technological breakthrough will be sustained on memristive, ferroelectric and floating gate two- and three-terminal devices based on 2D materials.

ENERGIZE seeks to enhance the hardware for the implementation of artificial neural networks through chiplet-based multi-core in-memory computing technologies, providing also guidelines for evaluating and benchmarking 2D devices and circuits.

ENERGIZE will be conducted by interconnected research integrating partners of Europe and Korea, accelerating the advancement of neuromorphic technology.ENERGIZE will demonstrate: wafer-scale growth of 2D materials for neuromorphic devices, reliable fabrication and characterization processes of two- and three-terminal devices, development of arrays of 2D devices compatible with existing technologies, efficient inference and training of neural networks in crossbar arrays and standardized benchmarking methods for neuromorphic devices and circuits.

ENERGIZE offers a promising way to set a new paradigm in low-power edge computing, departing from conventional large energy consumption and high environmental impact von Neumann CMOS AI hardware.

ENERGIZE aims to achieve substantial progress in the design, material growth, fabrication, integration, and characterization of neuromorphic devices, arrays, circuits, and systems, which will be key in the future of edge AI computing.

We will contribute to guiding an ethically sensitive design by integrating social sciences and humanities by consulting.ENERGIZE will stimulate the emergence of a European – Korean collaborative network for responsible neuromorphic 2D material-based hardware.

This integrated approach is a blueprint that will transform societies and economies of the future and will help Europe and Korea to enhance their position in chip technology for AI and machine learning applications.