Artificial Intelligence in Materials Science: Applications of Machine Learning to Extraction of Physically Meaningful Information from Atomic Resolution Microscopy Imaging

Materials science is the cornerstone for technological development of the modern world that has been largely shaped by the advances in fabrication of semiconductor materials and devices. However, the Moore’s Law is expected to stop by 2025 due to reaching the limits of traditional transistor scaling. However, the classical approach has shown to be unable to keep up with the needs of materials manufacturing, requiring more than 20 years to move a material from discovery to market. To adapt materials fabrication to the needs of the 21st century, it is necessary to develop methods for much faster processing of experimental data and connecting the results to theory, with feedback flow in both directions. However, state-of-the-art analysis remains selective and manual, prone to human error and unable to handle large quantities of data generated by modern equipment. Recent advances in scanning transmission electron and scanning tunneling microscopies have allowed imaging and manipulation of materials on the atomic level, and these capabilities require development of automated, robust, reproducible methods.Artificial intelligence and machine learning have dealt with similar issues in applications to image and speech recognition, autonomous vehicles, and other projects that are beginning to change the world around us. However, materials science faces significant challenges preventing direct application of the such models without taking physical constraints and domain expertise into account.Atomic resolution imaging can generate data that can lead to better understanding of materials and their properties through using artificial intelligence methods. Machine learning, in particular combinations of deep learning and probabilistic modeling, can learn to recognize physical features in imaging, making this process automated and speeding up characterization. By incorporating the knowledge from theory and simulations with such frameworks, it is possible to create the foundation for the automated atomic scale manufacturing.