Non-volatile programmable components for the superconducting computer

点击翻译按钮获取中文摘要

The dissipation of heat in traditional silicon (CMOS) based electronics is a major source of inefficiency and environmental impact. Superconductors are, by nature, dissipationless. Building a new digital computational scheme based on superconductors offers huge potential gains in energy efficiency, which could be applied to large scale needs such as supercomputers and data centres to lower the overall carbon footprint of the ICT ecosystem. Digital superconducting computers also have important niche applications such as in the control and read-out of qubits. The fundamental building block of the superconducting computer is the Josephson junction, which is proposed as a replacement for the CMOS transistor. Computing via logic circuits based on Josephson junctions is not only more energy efficient, but also faster than CMOS technologies. The largest remaining problem is the lagging development of low-temperature memory. Traditional magnetic and CMOS memories are not well suited for low-temperature operation. To achieve the promised efficiency increases of these computers requires a new type of low-temperature memory architecture. Our proposal here aims to combine superconducting circuits with magnetic memory elements, such as spin-valves, where the information is encoded into the magnetic state of the device but writing and read-out is achieved by the superconducting elements of the circuit for maximum energy efficiency and compatibility with the rest of the computer. Traditionally considered competing phenomena, when artificially juxtaposed a wealth of new physics at the interface between superconductors and ferromagnets emerges. It is possible to use these proximity effects to create new, highly energy efficient devices in two ways. The first exploits the natural competition between the phenomena, where information is stored by a "zero - pi" ground-state phase difference across a Josephson junction containing a ferromagnetic barrier. The second exploits a unique synergy, found in the form of the (equal spin) triplet Cooper pair. This new triplet Cooper pair is not broken by the exchange field of a ferromagnet, providing a dissipationless source of spin polarized current for use in novel spintronic devices. We plan to fully exploit these new physics in our research programme to open the potential novelty of our devices. The UK risks being left behind in the field of superconducting computing by large US research efforts such as the IARPA C3 programme. This programme will take advantage of our existing expertise and collaboration with leading US groups to develop the promising application of cryogenic memory.