CPM Seminar

Principles of and progress in building atomic electronic machines for classical and quantum circuitry: they really work

Bob Wolkow

University of Alberta

Our atomic silicon electronic machines are now working in a primitive way. Prospect for scaling to useful, commercially viable machines within a few years appear good.

Decades of development of the atomic silicon quantum dot provides the foundation for this. I will describe that ASiQD and the tricks developed to make it. While many in the community have been making silicon dangling bonds on hydrogen atom-terminated silicon surfaces for many years, it was only recently that it was recognized that this DB is quite special, and, methods to make truly perfect patterns of ASiQDs were discovered. In addition to existing on the very most attractive substrate, Si, and therefore the opportunity to be meshed with conventional CMOS circuitry, the silicon DB, and ensembles made of it, are in the band gap of the substrate silicon and as a result remain electrically isolated from the bulk conduction and valence bands allowing an all new generation of silicon circuitry to be built that is capable of extreme low power consumption and extreme high speed, as well as both classical and quantum functions.

Proof of concept binary circuit operation will be shown. The material embodiment of a spin Hamiltonian will be shown. Indications that this, with small changes, will embody a hardware Gibbs sampler will be discussed. Such a device could revolutionize neural network training efficiency.