The future of memristive technologies

The future of memristive technologies 

Funder: Royal Academy of Engineering

Lead partner: UCL Electronic and Electrical Engineering

Other partners: Intrinsic

Lead academic: Dr Adnan Mehonic

Project amount: £225,000

Research themes: Semiconductors

Project period: 1 Sep 2022 - 31 Aug 2027

Project description: Modern computing systems consume far too much energy. This is a key societal challenge, as powerful computing systems are indispensable in a wide range of rapidly growing data-centric technologies (e.g. AI, IoT, medicine, security, health tracking). The central technological challenge holding back many real-world applications (especially where energy resources are limited) is the memory bottleneck - the inability to accelerate the data movement between processor and memory and make it more energy-efficient.

This exciting research programme will solve the problem by developing SiOx memristors that will: (a) offer embeddability at the advanced processing nodes (<20nm; current nonvolatile memory (NVM), e.g. Flash, cannot be easily integrated into the processing chips fabricated with nodes <40/28nm); (b) achieve performance (speed, energy-efficiency) >100X better than NOR Flash; (c) minimise the fabrication costs. More fundamentally, developed memristors will enable novel analogue and neuromorphic processing to transform low power computing in edge technologies. I will work with Intrinsic and explore three outstanding opportunities: 1. Develop hugely improved memristive NVM for the next generation of microcontrollers. 2. Develop analogue memory for extremely energy-efficient AI chips. 3. Develop the key enablers for neuromorphic spiking-based systems.

Dr Adnan Mahonic's research profile