Technical validation device, both atomic machine and electronic circuit

The University of Alberta and Quantum Silicon Inc. are advancing in the field of atomic electronics. According to the University of Alberta, atomic models that control electrons could reveal the potential of "the most eco-friendly electronics," to reach applications such as neural networks for neurons. automatic learning.

"Atoms are a lot like chairs on which electrons sit," says Robert Wolkow, physics professor and principal investigator of the project. "To the extent that we can influence conversations at a dinner party by controlling the grouping of allocated chairs and seats, controlling the placement of individual atoms and electrons can affect the conversations between electronic components. "

Structures under atomic control have already been seen; However, until now, it was not possible to create custom models to develop electronic devices, Wolkow points out.

According to the University of Alberta, two major limitations precluded practical electronic applications. "The atoms would stay in place only at the cryogenic temperature and could be easily obtained only on metal surfaces that would not be technologically useful."

In a proof-of-principle device, the research team created a half-atomic, half-electronic system and overcame its two main design challenges. The device can also be enlarged because it can be configured on silicon surfaces.

"It's the icing on a cake we've been cooking for about 20 years," says Wolkow. "We have recently perfected the configuration of the silicon atoms, then we have learned through the machine to take over, thus relieving the long-suffering scientists.Now we have released the electrons for them to follow their nature – they can not leave the construction site we have created, but they can run freely and play with the other electrons out there.The positions of the electrons arrive, surprisingly, are the result of useful calculations. "

The team has started on larger machines that will simulate neural networks. Conventional systems consist of software-driven transistors. However, the atomic machine will "spontaneously display the relative energy stability of its bit patterns". Thus, the neural network can be driven "faster and more accurately".

Full details of the research can be found in the article entitled "Initiation and Monitoring of the Evolution of Simple Electrons in Structures Defined by Atoms", published in Letters of physical examination.