Researchers are using AI to assemble down to the atomic level

May.29.2018

Author :Justin Brunnette

Category: IT News

Researchers are using AI to assemble down to the atomic level

There has been quite a bit of hype over the last couple years in the realm of manufacturing since consumer 3D printers have hit the markets. Some have even speculated that someday we may even have a universal 3D printer that uses single atoms as its “ink” to potentially make anything imaginable. Although still a pipedream, the realm of engineering on the atomic level has seen some development with the help of AI.
 
Robert Wolkow, a researcher at the University of Alberta, has spent much of his career developing manufacturing at the atomic level and has been working to improve the scanning probe microscope. Known as the sharpest object ever made, it has a needle tip that is sharpened to the size of a single atom wide.
 
Researchers would use this tool much like a paintbrush to manipulate individual molecules and were able to achieve some valuable groundbreaking works. Once such example of their works is the construction of the smallest transistor in the world out of a single atom. Dr. Wolkow’s new plan is to use this tool to create computer chips with a circuit design to store information by manipulating atoms in the circuit.
 
The chips in our computers now have transistors that either hold electrons or release them to represent binary data (1 or 0). The moving of these electrons uses a lot of energy just to record information. But with the new design, there would be atoms in the circuit that has an electron and others with no electrons. The atoms would be manipulated to share or move electrons between them to which would be used to represent binary data. This arrangement uses significantly less energy since it would only be required to rearrange electrons rather than moving entire currents of electrons.
 
The researcher have the knowhow of manufacturing chips of this design using the scanning probe microscope for quite some time. The difficulty was using the scanning probe microscope as it is very difficult to use to manipulate single atoms and requires much time and patience. Researchers have found that they are able to streamline this process by combining machine learning with a method of correcting an atom’s position in a technique called “atomic whiteout.”
 
This is essentially the usage of an AI to sharpen the tip of the probe and correcting errors in atom’s positions. Based on machine learning, this software scans and the condition of the probe and reconditions the tip. They trained their AI to recognize the specific malformation in the structure in surface dangling bonds of various atoms.
 
This technique could potentially greatly increases the economic viability of manufacturing on an atom by atom scale. Dr. Wolkow’s goal is to be able to produce millions of the previously mentioned chip designs each year. The benefits of this research may be benefited by may other applications. European researchers, for example, were able to construct a computer memory from single atoms. Australian researchers have made quantum computers components with the placement of phosphorus atoms on silicon. Even if these were not able to be economically optimized, the research has made it easier for others in the field to use the scanning probe microscopes.
 
Original Article: https://www.wired.com/story/scientists-are-using-ai-to-painstakingly-assemble-single-atoms/