Research Group: Center for Condensed Matter Physics Full-time Project: yes
This project has been supported by the Faculty for CONACyT and/or Colfuturo funding. If you wish to discuss this or be considered for another funding route, please contact the supervisor [j.mol@qmul.ac.uk].
Computers, data centres and networks consume about 10% of the world's electricity. No area of energy consumption is further from its thermodynamic limit than computing. The minimum energy required to reset a bit is kBT ln2 [1,2]; this is known as the Landauer limit. At room temperature kBT ln2 = 3 zJ. The smallest silicon transistors today, made by only Intel, TSMC and Samsung, dissipate about 1 fJ per operation. This is 300,000 times the Landauer limit. Of course, there are other considerations, but the point is that there is plenty of scope for doing better. This project will explore routes towards low-power computing in molecular and nanoscale junctions [3,4] and networks using reversible and non-von Neumann type architectures [5]. Moreover, since a significant fraction of the energy consumed in conventional computing is lost to the environment in the form of high entropy ‘waste’ heat, technologies that can scavenge part of this waste heat could further reduce energy consumption. Since state-of-the art microprocessors generate tens of watts of heat in a spatially and temporally highly non-uniform fashion there is a significant need for site-specific and on- demand on-chip cooling that could be achieved by integrated thermoelectric nanodevices [6,7] that will be developed in this project.
[1] Pyurbeeva & Mol, Entropy 23, 640 (2021)
[2] Pyurbeeva et al., arXiv:2109.06741 (2021)
[3] Mol et al., Proc. Natl. Acad. Sci. 108, 13969–13972 (2011)
[4] Thomas et al., Nat. Commun. 10, 4628 (2019)
[5] Li et al., Sci. Rep. 6, 33686 (2016)
[6] Gehring et al., Nano Lett. 17, 7055–7061 (2017)
[7] Harzheim et al., Phys. Rev. Res. 2, 013140 (2020)
SPCS Academics: Dr. Jan Mol