Active Computing & AI Brain & Nervous System

Efficient and Robust Oxide Switching

Summary

Original abstract (not yet simplified)

We are at the beginning of a Data Age. Data is exploding. In 2016, 90% of the world’s data ever created was in the two previous years. AI and data analytics are further increasing the growth. The power demand is huge and growing. Within a few years some developed countries will not have sufficient power to sustain the growth. The...

View original technical description
We are at the beginning of a Data Age. Data is exploding. In 2016, 90% of the world’s data ever created was in the two previous years. AI and data analytics are further increasing the growth. The power demand is huge and growing. Within a few years some developed countries will not have sufficient power to sustain the growth. The negative effects on the planet are serious. Non-volatile memory (NVM) technology (including memory and neuromorphic computing elements in a single device) could strongly help to solve the problem, giving two orders of magnitude power reduction and, by removing the data transfer bottleneck, increased speed. Oxide memristors have significant advantages over competing NVM technologies, particularly in terms of speed, cost and temperature stability. However, after more than a decade of intense effort, serious challenges remain in terms of scaling, uniformity and robustness. The challenges all relate to a lack of precise control of the materials. Completely new thinking in thin film materials engineering is needed.EROS provides this new thinking by designing and engineering new forms of nanostructured oxide films to give highly Efficient, Robust Oxide Switching in an ultra-dense, ultra-low power, reliable oxide memristor system, with potential to change the technology landscape in AI, IoT, and security. ‘Ideal’ films will first be designed, fabricated, and understood. These will direct the way to simple industry-platform devices. Stochastic effects will be eliminated by creating films with separate vertical nanoscale ionic and electron channels with highly controlled vacancy and electronic concentrations, allowing scaling to a few nm, in a forming-free system. Also, multifunctional hybrid structures will be developed to give robustness. Furthermore, ferroelectricity will be induced, allowing simpler and smaller devices. Confidence in the proposed approach comes from proof-of-concept model systems shown by the PI.

Related Research

Grants with similar aims, by meaning.

Atomically Thin Oxides for Ultralow Power Non-volatile Memory Technology
ECCS-EPSRC: Overcoming the Endurance Challenge in Energy-Efficient Atomic Memristors for AI and 6G Applications
Memristive Organometallic Devices formed from self-assembled multilayers (MemOD)
Metal oxide films for ultralow energy memristive devices
PMRAM: Initial chip design for a novel non-volatile low-power piezomagnetic random access memory to drive the future of IOT

Original classification

H2020

Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.