The New Frontier of Atomic-level Manufacturing [Flash Forward Fridays]

Flash Forward Fridays

For the past few decades, technology has been evolving at a blink-and-you’ll-miss-it rate. In this biweekly column, we’ll be peeling back the curtain of the present, and exploring the developing technologies that may soon become the standard in the not-so distant future.

Peruse any number of electronics manufacturing websites, and each one will advertise their intensely tight tolerances. In a world where precision is becoming increasingly competitive, the manufacturing of electronics has become a race measured in microns and nanometers.

To give you some perspective on the size of this scale, consider that a single strand of human hair measures 100,000 nanometers; so these tolerances are catering to manufacturing in a realm that exists within a hair’s breadth.

Within this world of microscopic precision, manufacturing on the atomic level would be a crowning achievement. Imagine a level of accuracy in which control can be pinpointed to organizing and constructing at an atomic level during the manufacturing process. Such a capability would launch the manufacturing world into a new, revolutionary frontier.

Unsurprisingly, assembling atoms is exceedingly difficult; singular atoms are tricky to manipulate. However, in 1981, Dr. Gerd Binnig and Dr. Heinrich Rohrer, two Swiss scientists working with IBM in Zurich, invented the scanning tunneling microscope (STM), an instrument which was capable of capturing surface images at the atomic level.

This invention led to scanning probe microscopy (SPM), which utilizes a probe tip that tapers to the size of a single atom, making it possible to manipulate single atoms.

Recently, scientists at the University of Alberta’s Department of Physics combined this groundbreaking technology with deep-learning artificial intelligence to envision a new mode of electronics manufacturing. The research paper, titled Autonomous Atomic Scale Manufacturing through Machine Learning, was published in February 2019.

Researchers on the project “trained a convolutional neural network to identify and locate surface features in scanning tunneling microscope images of the technologically relevant hydrogen-terminated silicon surface,” the report explains. "Once the positions and types of surface features are determined, the predefined atomic structures are patterned in a defect-free area.”

Basically, the idea is to fabricate computer chips by “printing” super-tiny circuits directly onto silicon chips one atom at a time. If implemented on a commercial scale, this method of atomic assembly would not only push the electronics manufacturing industry way beyond the boundaries of its current methods, it would also spark a new revolution for computing.