Using a combination of high-resolution X-rays, time-recovery algorithms, and machine learning, Cornell University researchers have revealed complex nanotextures (Machine learning enhances X-ray imaging of nanotextures) in thin films, giving scientists a new, sophisticated way to examine those with power for large scale assembly and microelectronics among others application.
Scientists are particularly interested in nanotextures that are not uniformly distributed in thin films, because they can provide new properties. The best way to study nanotextures is to visualize them directly; a challenge that often requires difficult electron microscopy and does not preserve the sample.
A new imaging technique described on July 6 in the Proceedings of the National Academy of Sciences overcomes these challenges by using time-reversal and machine learning to interpret conventionally collected X-ray diffraction data — such as that produced at the Cornell High Energy Synchrotron Source, where the data was collected for the study – and the real space visualization of the objects in the nanoscale.
The use of X-ray diffraction allows scientists to access the system and allows other images to be taken, Andrej Singer, assistant professor of material science and engineering at the David Croll Sesquicentennial Faculty. A member of Cornell Engineering, who did the research. with doctoral student Ziming Shao.
“Remembering a large area is important because it represents the actual state of the property,” Singer said. “Nanotexture measured by field probes can depend on the choice of probe point.”
Another advantage of the new method is that it does not need to separate the sample, which allows a deeper study of important things like introducing light to see how the things develop.
“This method can be easily applied to study the situation or operando dynamics,” Shao said. “For example, we plan to use the system to study how the structure changes in picoseconds after excitation with short laser pulses, which may lead to new ideas for terahertz technology in the future.”
The technique was tested on two thin films, the first with a well-known nanotexture used to display the imaging results. By testing the second thin film – Mott insulator and the physics associated with superconductivity – the researchers discovered a new morphology that had not been seen in the material before – pressure-induced nanopatterns that develop simultaneously when it freezes to cryogenic temperatures.
“These images are generated without prior knowledge,” Shao said, “and can set new foundations and explain new physical concepts in time-generation structures, molecular dynamics simulations, and mechanical calculations.”
Source: Cornell University
