Google DeepMind, the AI research laboratory, has announced the discovery of millions of new crystals using a tool called Graph Networks for Materials Exploration (GNoME).
Existing crystal structures are utilized for all types of purposes from computer chips and solar chips.
Millions of crystals are theoretically possible but have remained undiscovered until now.
Finding new and stable crystalline structures is a little like seeking a needle in a haystack, and has historically required scientists in labs to make small structural changes to known crystals.
GNoME was able to predict the existence of 2.2 million new crystal structures, and more importantly to predict which of these theoretical structures would be stable.
The result was 380,000 new stable materials that can now be synthesized.
For context, previous scientists had been able to discover 20,000 crystals.
Recently, using computational analysis scientists were able to predict an additional 28,000 structures.
DeepMind’s discovery of 380,000 new crystals has been shared with scientists around the world through The Materials Project, a multi-institution, multi-national effort to compute the properties of all inorganic materials and provide the data and associated analysis algorithms for every materials researcher free of charge.
The discovery was also published in the scientific journal, Nature.
The resulting uses of these newly discovered crystals remain to be seen, but scientists have already synthesized more than 700 of the materials in the lab.
DeepMind states that it has discovered over 50,000 new types of graphene which is often used for producing superconductors.
In addition, GRaPH has predicted 528 new ion conductors that can be used to create newer and longer-lasting lithium batteries.