Automated Spectroscopy via Edge Computing for Smarter Instrumentation
An automated scanning probe microscopy platform has been developed to quickly identify high-response regions in samples.
Significance and Impact
This approach utilizes algorithms running on edge devices and increases efficiency while enabling new types of spectroscopic experiments that were previously not feasible.
– A Bayesian optimization algorithm run on a GPU server connected to the microscope facilitates smart sampling of the next batch of points to identify regions of maximal piezoelectric response in a ferroelectric sample.
– Most high response regions were found within just 30% of the time needed for the full data acquisition.
– Incorporating prior knowledge in the form of high-resolution imaging into this method is shown to further improve efficiency.
R. K. Vasudevan, K. P. Kelley, J. Hinkle, H. Funakubo, S. Jesse, S. V. Kalinin, and M. Ziatdinov, "Autonomous Experiments in Scanning Probe Microscopy and Spectroscopy: Choosing Where to Explore Polarization Dynamics in Ferroelectrics," ACS Nano (2021). DOI: 10.1021/acsnano.0c10239