AI discovers new nanostructures – Zoo House News
Scientists at the US Department of Energy’s (DOE) Brookhaven National Laboratory have successfully demonstrated that autonomous methods can discover new materials. The technique, based on artificial intelligence (AI), led to the discovery of three new nanostructures, including a unique nanoscale “ladder”. The research results were published today in Science Advances.
The newly discovered structures were formed through a process called self-assembly, in which a material’s molecules organize themselves into unique patterns. Scientists at the Brookhaven Center for Functional Nanomaterials (CFN) are experts at directing the self-assembly process and templates materials to form desirable assemblies for applications in microelectronics, catalysis, and more. Their discovery of nanoconductors and other new structures further expands the scope of self-assembly.
“Self-assembly can be used as a technique for nanopatterning, which is driving advances in microelectronics and computing hardware,” said CFN scientist and co-author Gregory Doerk. “These technologies are always pushing for higher resolution with smaller nanopatterns. You can get really small and tightly controlled features from self-assembling materials, but they don’t necessarily obey the rules that we set for circuits, for example. By directing self-assembly using a template, we can form patterns that are more useful.”
Staff at CFN, a user facility of the DOE Office of Science, plan to build a library of self-assembled nanopattern types to expand their applications. In previous studies, they have shown that new types of patterns are possible by mixing two self-assembling materials together.
“The fact that we can now create a ladder structure that no one has dreamed of before is amazing,” said CFN group leader and co-author Kevin Yager. “Conventional self-assembly can only form relatively simple structures such as cylinders, plates and spheres. But by mixing two materials and using just the right chemical lattice, we found that entirely new structures are possible.”
Mixing self-assembling materials has allowed CFN scientists to uncover unique structures, but it has also created new challenges. Since there are many more parameters to control in the self-assembly process, finding the right combination of parameters to create new and useful structures is a race against time. To speed up their research, CFN scientists used a new AI capability: autonomous experimentation.
In collaboration with the Center for Advanced Mathematics for Energy Research Applications (CAMERA) at DOE’s Lawrence Berkeley National Laboratory, Brookhaven scientists at CFN, and the National Synchrotron Light Source II (NSLS-II), another user facility of the DOE Office of Science im Brookhaven Lab, have developed an AI framework that can autonomously define and run all steps of an experiment. CAMERA’s gpCAM algorithm drives the framework’s autonomous decision making. The latest research is the team’s first successful demonstration of the algorithm’s ability to discover new materials.
“gpCAM is a flexible algorithm and software for autonomous experimentation,” said Berkeley Lab scientist and co-author Marcus Noack. “It was used particularly cleverly in this study to autonomously explore various features of the model.”
“With the help of our colleagues at the Berkeley Lab, we had this software and methodology ready to go, and now we’ve used it successfully to discover new materials,” Yager said. “We’ve now learned enough about autonomous science that we can take a materials problem and turn it into an autonomous problem fairly easily.”
To speed up material discovery with their new algorithm, the team first developed a complex sample with a range of properties for analysis. Researchers fabricated the sample using the CFN nanofabrication facility and performed the self-assembly in the CFN materials synthesis facility.
“One way of doing old-school materials science is to synthesize a sample, measure it, learn from it, and then go back and make another sample and keep repeating that process,” Yager said. “Instead, we made a sample that has a gradient of all the parameters we are interested in. So this single sample is a huge collection of many different material structures.”
The team then took the sample to NSLS-II, which produces ultra-bright X-rays to study the material’s structure. CFN operates three experimental stations in partnership with NSLS-II, one of which was used in this study, the Beamline Soft Matter Interfaces (SMI).
“One of the strengths of the SMI beamline is its ability to focus the X-ray beam on the sample to within microns,” said NSLS-II scientist and co-author Masa Fukuto. “By analyzing how these microbeam X-rays are scattered by the material, we learn about the local structure of the material at the illuminated point. Measurements at many different points can then show how the local structure changes across the gradient sample. In this work, we let the AI algorithm choose which spot to measure next on the fly to maximize the value of each measurement.
When the sample was measured on the SMI beamline, the algorithm created a model of the material’s numerous and varied structures without human intervention. The model updated with each subsequent X-ray measurement, making each measurement more insightful and accurate.
Within hours, the algorithm had identified three key areas in the complex sample for CFN researchers to examine more closely. They used the CFN electron microscopy facility to image these key areas in exquisite detail, revealing the rails and rungs of a nanoscale ladder, among other novel features.
From start to finish, the experiment took about six hours. The researchers estimate it would have taken them about a month to make this discovery using conventional methods.
“Autonomous methods can speed up discovery tremendously,” Yager said. “It’s essentially a ‘narrowing’ of science’s usual discovery loop so that we move more quickly between hypotheses and measurements. Beyond sheer speed, however, autonomous methods expand the scope of what we can study, meaning we can tackle more challenging scientific problems. “
“In the future we want to investigate the complex interaction of several parameters. We performed simulations with the CFN computer cluster, which confirmed our experimental results, but they also indicated that other parameters such as film thickness may also play an important role,” said Dörk.
The team is actively applying their autonomous research methodology to even more challenging material discovery problems in self-assembly, as well as to other classes of materials. Autonomous discovery methods are adaptable and can be applied to almost any research problem.
“We are now making these methods available to the broad community of users who come to CFN and NSLS-II to conduct experiments,” Yager said. “Anyone can work with us to accelerate their exploration of materials. We expect this to enable a multitude of new discoveries in the years to come, including in national priority areas such as clean energy and microelectronics.”
This research was supported by the DOE Office of Science.