Original source: Academic Headlines
Image source: Generated by Unbounded AI
In just 17 days, artificial intelligence (AI) has created 41 new materials on its own, more than two per day.
In contrast, it can take months of trial and error for human scientists to create a new material.
Today, the AI lab, called A-Lab, is featured in the authoritative scientific journal Nature. **
According to the introduction, **A-Lab is a laboratory where AI-guided robots make new materials, which can quickly discover new materials with minimal human intervention, which can help identify and fast-track materials in multiple research fields, including batteries, energy storage, solar cells, fuel cells, etc.
It is worth mentioning that in a test mission, A-Lab successfully synthesized 41 of the 58 predicted materials, with a success rate of 71%.
The test data comes from the Materials Project, a Berkeley Lab open-access database, and the Graph Networks for Materials Exploration (GNoME) deep learning tool developed by Google DeepMind.
Also today, Google DeepMind’s GNoME is featured in Nature, contributing nearly 400,000 new compounds to the Materials Project, the largest addition of new structural stability data by a single team since the project’s inception, greatly increasing the open access resources scientists can use to invent new materials for future technologies.
Kristin Persson, founder and director of the Materials Project at Berkeley Lab and professor at the University of California, Berkeley, said, "To solve global environmental and climate challenges, we must create new materials. With material innovation, we can develop recyclable plastics, harness waste energy, make better batteries, and build cheaper, longer-lasting solar panels, among other things**. ”
With AI, it is faster to manufacture and test new materials
The development of new technologies often requires new materials. However, manufacturing a material is not an easy task.
Scientists have calculated hundreds of thousands of new materials, but testing whether they can be made in reality is a slow process. It takes a long time for a material to go from calculation to commercialization. It must have the right attributes, be able to work in the device, be scalable, and have the right cost efficiency and performance.
Today, thanks to supercomputers and simulations, researchers no longer have to blindly try to create material from scratch.
In this work, the Google DeepMind team trained GNoME using workflows and data developed by the Materials Project over a decade and improved the GNoME algorithm through active learning.
As a result, GNoME produced 2.2 million crystal structures, of which 380,000 were included in the Materials Project and predicted to be stable. These data include the arrangement of atoms of the material (crystal structure) and stability (formation energy).
The compound Ba₆Nb₇O₂₁ is one of the new materials calculated by GNoME and contains barium (blue), niobium (white) and oxygen (green).
According to the paper, GNoME has improved the accuracy of prediction of structural stability to more than 80%, and the accuracy of prediction of components to 33% per 100 trials (compared to 1% in previous work).
Ekin Dogus Cubuk, Head of the Materials Discovery Team at Google DeepMind, said: "We hope that the GNoME project will advance the research of inorganic crystals. More than 736 new materials discovered by GNoME have been validated by external researchers through independent physical experiments, proving that the discovery of our model can be achieved in the laboratory. ”
However, the research team also points out in the paper that there are still some open questions about GNoME in practical applications, including the dynamic stability caused by phase transitions, vibrational profiles, and configuration entropy caused by competing polymorphs, as well as a deeper understanding of the final synthesis capacity.
To create the new compounds predicted by the Materials Project, A-Lab’s AI created new formulations by studying scientific papers and tweaking them using active learning.
Gerd Ceder, a scientist at Berkeley Lab and UC Berkeley, principal investigator at A-Lab, said, "We’ve had a staggering 71 percent success rate, and we’ve found some ways to improve. We’ve proven that combining theory and data with automation yields incredible results. We can manufacture and test materials faster than ever before. ”
According to reports, with some small changes to the decision-making algorithm, this success rate can be increased to 74%, and if the computing technology is improved, the success rate can be further increased to 78%.
“Not only do we want to make the data we produce free and usable to accelerate material design around the world, but we also want to teach the world what computers can do for people,” Persson said. They can scan a wide range of new compounds and properties more efficiently and quickly than experiments alone. ”
With the help of the likes of A-Lab and GNoME, scientists can focus on promising materials for future technologies, such as lighter alloys that improve fuel economy in automobiles, more efficient solar cells that improve the efficiency of renewable energy, or faster transistors in next-generation computers.
Proven application potential
Currently, the Materials Project is processing more Google DeepMind’s compounds and adding them to an online database. The new data will be made available to researchers free of charge and will also be fed into projects such as A-Lab, which collaborates with the Materials Project.
Figure: Structures of 12 compounds in the Materials Project database.
Over the past decade, researchers have experimentally confirmed the usefulness of new materials in a number of areas, based on clues from Materials Project data. Some of them have shown application potential, such as:
- In carbon capture (extraction of carbon dioxide from the atmosphere)
- As a photocatalyst (a material that accelerates chemical reactions under the action of light, which can be used to break down pollutants or produce hydrogen)
- As a thermoelectric material (a material that helps to harness waste heat and convert it into electrical energy)
- As a transparent conductor (can be used for solar cells, touch screens, or LEDs)
Of course, finding these potential materials is just one of the many steps to solve some of the major technological challenges facing humanity.
**In addition to the above two researches, AI has made many breakthroughs in the discovery and synthesis of new materials in recent years. **
In 2020, a multi-agency research team, including the National Institute of Standards and Technology (NIST), developed an AI algorithm called CAMEO that autonomously discovered a potentially useful new material without additional training from scientists.
Figure | CAMEO’s process of finding new materials in a closed-loop operation (Source: NIST)
That same year, researchers from North Carolina State University and the University at Buffalo developed a technology called “Artificial Chemist,” which combines AI and automated systems that perform chemical reactions to accelerate R&D and production of new chemical materials needed for business.
In 2022, nanoengineers at the University of California, San Diego’s School of Engineering developed an AI algorithm, M3GNet, that can predict the structural and dynamic properties of any material, whether existing or new, almost instantaneously. Researchers can use it to find safer, higher energy-density electrodes and electrolytes for rechargeable lithium-ion batteries.
Figure | Schematic diagram of multibody diagram potential energy and main calculation modules (Source: University of California, San Diego)
In March, a study published in Nature Synthesis envisioned a future of accelerated materials science driven by the co-development of combinatorial synthesis and AI technologies. To evaluate the applicability of synthesis techniques to specific experimental workflows, the researchers established a set of ten metrics covering synthesis speed, scalability, range, and synthesis quality, and summarized some selective combinatorial synthesis techniques in the context of these metrics.
**As the foundation and forerunner of high technology, new materials have a wide range of applications, and it has become the most important and promising field in the 21st century together with information technology and biotechnology. **
In the future, with breakthroughs in technologies such as AI, scientists will be expected to focus on materials that are more promising in future technologies, such as lighter alloys that improve fuel economy in automobiles, more efficient solar cells that promote renewable energy, and faster transistors that will play a role in the next generation of computers.
Reference Links:
