Abstract

Research direction: Whilst a holistic approach to sustainable ICT involving hardware, software and usage is required, energy efficiency of the hardware is at the core of the problem, underpinning each of the other aspects. Digital electronics is seemingly utterly dominated by Si CMOS, with an entrenched position that results from decades of investment in research and development, especially scaling of transistors to dimensions that compete with research nanotechnology devices, making Si CMOS’ position seem unassailable. That may well be the case for logic (processors), but not so for memories, which are often overlooked in Moore’s Law (and beyond) discussions. In fact, the size of the 2019 memory market ($106bn) was almost identical to that of the logic market ($1087bn), and, in contrast to the monolith of Si CMOS technology that logic is, there are four forms of conventional (Si) memory, SRAM, DRAM, NAND flash and NOR flash, only one of which (SRAM) has CMOS as its fundamental basis. Furthermore, there is already vibrant research activity into new forms of memory, usually called emerging memories, that include phase change RAM, resistive RAM, and spin-transfer torque magnetic RAM. These emerging memories aspire to combine the non-volatility of flash with the performance of DRAM, and some are available commercially, e.g. Optane™. There is clearly scope for innovative memory research, and realistic prospects of actual commercialisation and impact.

Specific expertise: We have developed a compact, charge-based memory, ULTRARAM™, which exploits a triple-barrier resonant tunnelling structure [1] to deliver non-volatility in combination with ultra-low switching energy (100 time lower than DRAM per unit area) and intrinsic switching speeds that are predicted to match SRAM. Room temperature operation of single devices has been reported [2], principles of operation simulated [3] and array architecture developed [3]. Small (4-bit) arrays are currently under production. We are looking for partners to work with us on the development of large arrays of scaled devices on Si substrates, and to explore societal and use implications of the deployment of ultra-efficient non-volatile RAM in the smallest Internet of Things devices through to large-scale (cloud-based) storage and computing.

[1] Hayne, M., 2019. Electronic Memory Devices, US10243086B2.
[2] Tizno, O., Marshall, A.R.J., Fernández-Delgado, N., Herrer, M., Molina S.I., & Hayne, M., 2019. Room-temperature Operation of Low-voltage, Non-volatile, Compound-semiconductor Memory Cells, Scientific Reports 9, 8950.
[3] Lane, D., & Hayne, M., 2020. Simulations of Ultralow-Power Nonvolatile Cells for Random-Access Memory, IEEE Trans. Electron Devices 67, pp 474-480.