April 20, 2024
A.I

Google scientists discovered 380,000 new materials using artificial intelligence

Robot holding futuristic crystal

The Materials Project, an open access database for new materials, is revolutionizing the way researchers discover and develop materials for future technologies, with Google DeepMind contributing 400,000 new compounds. This synergy of artificial intelligence, supercomputing and experimental data accelerates the creation of materials for applications such as renewable energy, efficient electronics and environmental solutions. (Artist’s concept). Credit: SciTechDaily.com

Expanding the open access resource is critical for scientists to develop novel materials for future technologies.

New technological advances often require the development of novel materials, and thanks to supercomputers and advanced simulations, researchers can avoid the time-consuming and often inefficient trial-and-error process.

He Materials Project, an open-access database founded at the Department of Energy’s Lawrence Berkeley National Laboratory (Berkeley Lab) in 2011, calculates the properties of known and predicted materials. Researchers can focus on promising materials for future technologies: think lighter alloys that improve fuel economy in cars, more efficient solar cells to power renewable energy, or faster transistors for the next generation of computers.

Now, Google DeepMind (Google’s artificial intelligence lab) is contributing nearly 400,000 new compounds to the Materials Project, expanding the amount of information researchers can draw on. The data set includes how a material’s atoms are arranged (the crystal structure) and how stable it is (formation energy).


The Materials Project can visualize the atomic structure of materials. This compound (Ba₆Nb₇O₂₁) is one of the new materials calculated by GNoME. Contains barium (blue), niobium (white) and oxygen (green). Credit: Materials Project/Berkeley Lab

“We have to create new materials if we want to address global environmental and climate challenges,” said Kristin Persson, founder and director of the Materials Project at Berkeley Lab and a professor at UC Berkeley. “With innovation in materials, we can potentially develop recyclable plastics, harness waste energy, make better batteries, and build cheaper, longer-lasting solar panels, among many other things.”

GNoME’s role in materials discovery

To generate the new data, Google DeepMind developed a deep learning tool called Graph Networks for Materials Exploration, or GNoME. The researchers trained GNoME using workflows and data developed over a decade by the Materials Project and improved the GNoME algorithm through active learning. GNoME researchers ultimately produced 2.2 million crystal structures, including 380,000 that they are adding to the Materials Project and predict to be stable, making them potentially useful in future technologies. The new Google DeepMind results were recently published in the journal Nature.

a laboratory

Robots guided by artificial intelligence created more than 40 new materials planned by the Materials Project. The GNoME data was used as an additional check to determine if those predicted materials would be stable. Credit: Marilyn Sargent/Berkeley Laboratory

Some of the GNoME calculations were used along with data from the Materials Project to test A-Lab, a facility at Berkeley Lab where artificial intelligence guides robots in making new materials. The first A-Lab article, also published in Naturedemonstrated that the autonomous laboratory can rapidly discover novel materials with minimal human intervention.

During 17 days of independent operation, A-Lab successfully produced 41 new compounds out of 58 attempts, a rate of more than two new materials per day. By way of comparison, it can take a human researcher months of guesswork and experimentation to create a new material, if he ever reaches the desired material.

To create the new compounds predicted by the Materials Project, A-Lab’s AI created new recipes by reviewing scientific papers and using active learning to make adjustments. Data from the Materials Project and GNoME were used to evaluate the expected stability of the materials.

Grid project of crystalline structures from materials.

Berkeley Lab’s Materials Project gives researchers access to crucial information about various materials. This image shows the structures of 12 compounds in the Materials Project database. Credit: Jenny Nuss/Berkeley Laboratory

“We had this amazing 71% success rate and we already have some ways to improve it,” said Gerd Ceder, A-Lab principal investigator and scientist at Berkeley Lab and UC Berkeley. “We have shown that combining theory and data with automation produces incredible results. “We can manufacture and test materials faster than ever, and adding more data points to the Materials Project means we can make even smarter decisions.”

The impact and future of the materials project

The Materials Project is the world’s most widely used open-access repository of information on inorganic materials. The database contains millions of properties on hundreds of thousands of structures and molecules, information processed primarily at Berkeley Laboratory’s National Energy Research Scientific Computing Center. More than 400,000 people are registered as users of the site and, on average, more than four articles citing the Materials Project are published each day. Google DeepMind’s contribution is the largest addition of structural stability data by any group since the Materials Project began.

“We hope the GNoME project will boost research on inorganic crystals,” said Ekin Dogus Cubuk, Materials Discovery team leader at Google DeepMind. “External researchers have already verified more than 736 of the new GNoME materials through independent and simultaneous physics experiments, demonstrating that our model discoveries can be made in laboratories.”


This one-minute timelapse shows how people around the world use the Materials Project over the course of four hours. The data dashboard shows a rolling one-hour window of global Materials Project activity, including the number of requests, the country of users, and the most frequently queried material properties. Credit: Patrick Huck/Berkeley Laboratory

The Materials Project is now processing Google DeepMind compounds and adding them to the online database. The new data will be freely available to researchers and will also feed into projects such as A-Lab, which partner with the Materials Project.

“I’m really excited that people are using the work we’ve done to produce an unprecedented amount of information about materials,” said Persson, who is also director of Berkeley Lab’s Molecular Foundry. “This is what I set out to do with the Materials Project: not only make the data I produced free and available to accelerate materials design for the world, but also show the world what calculations can do for you . “They can scan large spaces for new compounds and properties more efficiently and quickly than experiments alone.”

NERSC

Many of the calculations for the Materials Project are performed on supercomputers at Berkeley Laboratory’s National Energy Research Scientific Computing Center. Credit: Thor Swift/Berkeley Lab

Following promising leads from Materials Project data over the past decade, researchers have experimentally confirmed useful properties in new materials in several areas. Some show potential for use:

  • in carbon capture (removing carbon dioxide from the atmosphere)
  • as photocatalysts (materials that speed up chemical reactions in response to light and could be used to break down pollutants or generate hydrogen)
  • as thermoelectrics (materials that could help take advantage of waste heat and convert it into electrical energy)
  • as transparent conductors (which could be useful in solar cells, touch screens or LEDs)

Of course, finding these potential materials is just one of many steps toward solving some of humanity’s great technological challenges.

“Making a material is not for the faint of heart,” Persson said. “It takes a long time to take a material from computing to commercialization. It has to have the right properties, work within devices, be able to scale, and have the right cost effectiveness and performance. “The goal of the Materials Project and facilities like A-Lab is to harness data, enable data-driven exploration, and ultimately provide companies with more viable opportunities to achieve the goal.”

Reference: “A self-contained laboratory for accelerated synthesis of novel materials” by Nathan J. Szymanski, Bernardus Rendy, Yuxing Fei, Rishi E. Kumar, Tanjin He, David Milsted, Matthew J. McDermott, Max Gallant, Ekin Dogus Cubuk, Amil Merchant, Haegyeom Kim, Anubhav Jain, Christopher J. Bartel, Kristin Persson, Yan Zeng, and Gerbrand Ceder, November 29, 2023. Nature.
DOI: 10.1038/s41586-023-06734-w

Leave a Reply

Your email address will not be published. Required fields are marked *