Image: A smart approach to microscopy and imaging developed at Oak Ridge National Laboratory could drive discoveries in materials for future technologies. Credits: Adam Malin/ORNL, U.S. Dept. of Energy
Microscopes are being taught to drive discoveries by researchers at the Department of Energy’s Oak Ridge National Laboratory, using an intuitive algorithm developed at the lab’s Center for Nanophase Materials Sciences that could lead to breakthroughs in new materials for energy technologies, sensing, and computing.
The method, described in Nature Machine Intelligence, combines physics and machine learning to automate microscopy experiments aimed at determining the functional features of materials at the nanoscale.
“There are so many potential materials, some of which we cannot study at all with conventional tools, that need more efficient and systematic approaches to design and synthesize,” said Maxim Ziatdinov of ORNL’s Computational Sciences and Engineering Division and the CNMS. “We can use smart automation to access unexplored materials as well as create a shareable, reproducible path to discoveries that have not previously been possible.”
Functional materials, ranging from computers and solar cells to artificial muscles and shape-memory polymers, respond to stimuli such as heat or electricity and are designed to serve both common and new technology. Their distinct characteristics are linked to atomic structures and microstructures visible with sophisticated microscopy. However, developing effective strategies to pinpoint regions of interest where these traits appear and may be examined has been a difficulty.
Instruments use an atomically sharp probe to scan the surface of materials to map out the structure at the nanoscale scale — one billionth of a meter. They can also detect reactions to a variety of stimuli, allowing researchers to learn more about the underlying principles of polarization switching, electrochemical reactivity, plastic deformation, and quantum events. Today’s microscopes can scan a nanoscale square grid point by point, but the operation is time-consuming, with measurements taken over days for a single substance.
Former ORNL CNMS scientist and lead author Sergei Kalinin, now at the University of Tennessee, Knoxville says, “the interesting physical phenomena are often only manifested in a small number of spatial locations and tied to specific but unknown structural elements. While we typically have an idea of what will be the characteristic features of physical phenomena we aim to discover, pinpointing these regions of interest efficiently is a major bottleneck. Our goal is to teach microscopes to seek regions with interesting physics actively and in a manner much more efficient than performing a grid search.”
To tackle this difficulty, scientists have turned to machine learning and artificial intelligence, but traditional algorithms require massive, human-coded datasets and may not save time in the end.
The ORNL procedure blends human-based physical reasoning into machine learning approaches and starts with extremely tiny datasets — photos from less than 1% of the sample — for a more intelligent approach to automation. The algorithm chooses places of interest depending on what it learns throughout the experiment as well as external knowledge.
A method was presented utilizing scanning probe microscopy and applied to well-studied ferroelectric materials as a proof of concept. Ferroelectrics are reorientable surface charge functional materials that may be used in computation, actuation, and sensing applications. Scientists want to know how the quantity of energy or information these materials can store is related to the local domain structure that governs this attribute. The automated experiment determined which topological faults these values are optimal for.
“The takeaway is that the workflow was applied to material systems familiar to the research community and made a fundamental finding, something not previously known, very quickly — in this case, within a few hours. We wanted to move away from training computers exclusively on data from previous experiments and instead teach computers how to think like researchers and learn on the fly. Our approach is inspired by human intuition and recognizes that many material discoveries have been made through the trial and error of researchers who rely on their expertise and experience to guess where to look,” said Ziatdinov.
The technical difficulty of getting the algorithm to operate on an operating microscope at the CNMS was handled by ORNL’s Yongtao Liu.
“This is not an off-the-shelf capability, and a lot of work goes into connecting the hardware and software. We focused on scanning probe microscopy, but the setup can be applied to other experimental imaging and spectroscopy approaches accessible to the broader user community,” said Liu.