UK data centres could guzzle six times more electricity within a decade, driven by the AI boom, unless silicon transistors are replaced with far more efficient alternatives. This project tackles the core problem: silicon devices waste energy as heat. The team will build ultra-low-energy semiconductor devices from atom-thin 2D materials—starting with graphene, then adding other 2D semiconductors like hBN and gallium selenide. Unlike silicon, where electrons move through the bulk, electrons in 2D materials glide along surfaces, enabling faster, cooler operation. The researchers have already scaled up graphene production and spun out a company, Paragraf, whose sensors use 1000 times less power than silicon equivalents. If successful, these new devices—from simple diodes to novel Dirac-source transistors—could cut data centre electricity demand by 100-fold. That would save billions of pounds in energy costs and help the UK meet net-zero targets while supporting the AI revolution. The project is applied, not fundamental: it aims directly at commercial prototype devices, backed by 23 industry partners contributing over £2 million. Success would give the UK a world lead in low-energy, high-speed electronics beyond silicon.
View original technical description
Our vision is to research and develop a new generation of two-dimensional (2D) semiconductors and ultra-low-energy semiconductor devices. Our research proposal addresses one of the most urgent, important and costly problems facing the UK: the huge energy required to power the data centres needed to train and run AI-enabled computing. John Pettigrew, CEO National Grid, warned at the Oxford Aurora Forum (March 2024): “The electricity demand from UK data centres is predicted to rise sixfold over the next 10 years.” The main reason data centres consume so much energy is that they contain millions of silicon devices such as transistors which are energy inefficient (our laptops get hot because of the energy-hungry silicon devices they contain). If we are to reduce the energy consumption of data centres, we must replace silicon devices with low-energy-consumption devices made from new materials. The front-runner materials for this are the atomically thin new class of materials called 2D materials. The first 2D material was graphene, a single layer of carbon atoms, the thinnest material in the world. Graphene is the best electrical conductor in the world. This is because the conduction electrons move along the surfaces of graphene, whereas in silicon they move through the bulk. Think of skating on ice: you can skate on the surface much faster than trying to travel through the ice. This results in high-speed graphene devices. IBM, Intel, and Samsung invested over $5 billion to manufacture graphene electronics; however, although graphene devices could be assembled by hand using the tiny graphene flakes then available, graphene could not be scaled up for high-volume manufacturing. This major technological drawback was removed when the Programme Lead's group, then at Cambridge University, invented a new way to scale up and made device-quality, large-area graphene. They then made a graphene sensor to measure magnetic fields and electric currents. A company, Paragraf, was set-up to manufacture graphene devices. Importantly, the power consumption of Paragraf’s sensors is 1000 times less than that of silicon sensors. This raises our expectations that other devices made from 2D materials will also have a very low power consumption, because the conduction electrons move along their surfaces. Our proposed next-generation, ultra-low-energy semiconductor devices aim to reduce the electricity consumption of data centres by 100 times. The power demand of UK data centres would then drop to a much more manageable value, which would save many £billions in electricity costs. Our research team from the universities of QMUL, Glasgow and Nottingham, supported by 23 partner industries contributing over £2 million to the project, will build upon our world lead in wafer-scale graphene electronic devices to produce more complex prototype devices using other 2D semiconductors (e.g., hBN, gallium selenide) at wafer-scale and integrate them, layer by layer with graphene. We will translate our unique scientific knowledge and expertise of 2D graphene electronic devices into new prototype semiconductor devices, ranging from simple diodes to nanometre-scale novel transistors, such as Dirac-source transistors suitable for integrated circuits for use in data centres. Our research will be revolutionary and give the UK a world lead in low-energy, high-speed electronics beyond silicon, enabling the UK to build a new electronics industry in 2D semiconductor devices.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know