Innovative materials are created atom by atom. These quantum LEGO bricks are not assembled by humans

This is best illustrated by a story from Asia, supported by representatives of the National University of Singapore. Thanks to their efforts, a tool was created that allows the creation of carbon-based quantum materials. They are arranged atom by atom by neural networks integrated with scanning electron microscopes.

Read also: Artificial intelligence will help target new materials to capture carbon dioxide

The publication dedicated to the achievements of Singaporean researchers was presented in Synthesis of nature. Apart from the obvious result of the research, which is the insight into the design of previously unknown materials with unique properties, there is another aspect that is also important. The idea is to show the usefulness of artificial intelligence in carrying out the most demanding tasks.

The focus of interest of the research team members was the so-called open-shell magnetic nanographene. Their properties mean that such materials could lead to a small revolution in the design of advanced electronics, for example quantum computers and other devices used to perform complex calculations.

Materials known as open-shell magnetic nanographs can be designed faster using a tool based on deep neural networks

An important component of recent research has been the concept of CARP (Atomic robotic probe chemist's intuition), where deep neural networks are trained using the knowledge provided by chemists. After undergoing training, this tool gains the ability to independently fabricate open-shell magnetic nanographene.

This sounds great in theory, but how did this solution work in practice? The authors of the aforementioned post say that it is very good. As they claim, the tool they designed is able to effectively absorb knowledge provided to it from the outside (by people), transform it so that machines “understand” it, and then design new materials that exhibit the desired properties.

Read also: Powerful laser in the hands of the Poles. His heartbeat is faster than the blink of an eye and works wonders

As if this were not enough, the CARP concept has been shown to have the ability to capture details that humans miss. The achievements achieved so far are impressive, but the authors do not intend to stop there. They explained that, as part of their further efforts, they would like to expand the CARP structure, which would translate into shifting the current laboratory surface synthesis process toward on-systems production, opening the door to practical applications.

The body of materials known to science is growing, but finding more of them using traditional methods is becoming more difficult. Advanced AI-based tools make this task easier. It allows thousands of different variables to be tested, incomparably reducing the time needed to identify new materials with useful properties.